Regions to Streams: Spatial and Temporal Variation in Stream Occupancy
Patterns of Coho Salmon (Oncorhynchus kisutch) on the Oregon Coast
Rebecca L. Flitcroft
Doctor of Philosophy Dissertation
19 October 2007
Fisheries Science
Oregon State University, Corvallis, Oregon
AN ABSTRACT OF THE DISSERTATION OF
Rebecca L. Flitcroft for the degree of Doctor of Philosophy in Fisheries Science
presented on 19 October 2007.
Title: Regions to Streams: Spatial and Temporal Variation in Stream Occupancy Patterns of Coho Salmon
(Oncorhynchus kisutch) on the Oregon Coast.
Abstract approved: ____________________________________________________________________
Gordon H. Reeves
Aquatic ecological investigation is expanding to encompass considerations of multiple scales across large
landscapes. Much of the analysis included in this work focuses specifically on coho salmon (Oncorhynchus
kisutch) in multiple subbasins on the Oregon coast. Coho salmon were chosen for an investigation of spatial
scales, network connections, and life history stages due to their broad distribution on the Oregon coast, and
abundant data describing their distribution, habitat needs, behavior, and survival. Chapter 2 introduces
dynamic network topology (DNT) as a framework for analysis and interpretation of aquatic obligate species.
DNT is based on the premise that in-stream habitats change in form and organization over time, and native
aquatic species are adapted to those changes through movement and life history diversity. Chapter 3 analyzes
juvenile coho salmon density and stream network occupancy at three spatial scales (site, patch, and subbasin).
The site scale analysis indicated that combining network and traditional in-stream habitat metrics (i.e.,
substrate and habitat juxtaposition variables) are most effective at describing juvenile coho salmon density.
Patch sizes of juvenile coho salmon were defined using variograms. Variogram shape indicated that a nested
spatial structure may be present in larger subbasins, indicating overlapping patterns of juvenile stream use. At
the subbasin scale, stream network occupancy by juvenile coho salmon was shown to vary over time within
subbasins, and appeared to increase or decrease similarly to the size of the adult spawning run. In chapter
3, two-tier Bayesian hierarchical models were applied to adult (subbasin and basin scales) and juvenile (site
and subbasin scales) coho salmon in an attempt to combine spatial scales that might be influential at each
life history stage. The best fitting adult model included the percent of large trees in the riparian zone at the
subbasin scale with mean annual precipitation at the basin scale. The best fitting juvenile model included
three variables, percent sand, stream order, and network distance to spawning habitat which mirrors the
result of modeling efforts in Chapter 3. Multiple spatial scales and the framework of a stream network
were informative at detecting patterns and interactions among scales and life history stages of coho salmon.
© Copyright by Rebecca L. Flitcroft
19 October 2007
All Rights Reserved
Regions to Streams: Spatial and Temporal Variation in Stream Occupancy
Patterns of Coho Salmon (Oncorhynchus kisutch) on the Oregon Coast
by
Rebecca L. Flitcroft
A DISSERTATION
submitted to
Oregon State University
in partial fulfillment of
the requirements for the
degree of
Doctor of Philosophy
Presented October 19, 2007
Commencement June 2008
Doctor of Philosophy dissertation of Rebecca L. Flitcroft presented on October 19, 2007.
APPROVED:
___________________________________________________________________
Major Professor, representing Fisheries Science
___________________________________________________________________
Head of the Department of Fisheries and Wildlife
___________________________________________________________________
Dean of Graduate School
I understand that my dissertation will become part of the permanent collection of Oregon State University libraries.
My signature below authorizes release of my dissertation to any reader upon request.
___________________________________________________________________
Rebecca L. Flitcroft, Author
ACKNOWLEDGMENTS
The work of this dissertation would not have been possible without the help of a large community of people and
organizations. I would first like to acknowledge the support and intellectual inspiration of my major professor, Dr.
Gordon Reeves. Gordie’s insight, experience, and enthusiasm for my research were a consistent source of strength for me.
My amazing committee members Lisa Ganio, Harold Batchelder, Alix Gitelman, and Robert Gresswell were instrumental
in all aspects of my research. I am indebted to each of them for taking the time to listen to my ideas and help guide my
analysis. I would also like to thank Kathryn Ronnenberg for her artistic interpretation of my figures. Kelly Burnett was
an irreplaceable resource to me in developing ideas, revising chapters, and being a kind ear to listen and support me on
the days when I needed that support the most. I would also like to acknowledge the GIS expertise of Kelly Christianson
who provided much needed assistance in compiling spatial datasets for Chapter 4.
I would also like to extend my thanks to the Mid-Coast Watershed Council for generously providing me with
the juvenile coho salmon snorkel datasets that were the core of all analyses in this dissertation. Their vision to collect
continuous datasets throughout an entire stream network made the work of this dissertation possible. I would also like to
thank Kim Jones of the Oregon Department of Fish and Wildlife’s Aquatic Inventories Project for his review of the early
drafts of this dissertation. Also, I would like to thank Kim for providing me with opportunities to explore habitat and
salmon relationships in the years before my dissertation research. Jeff Rodgers, Erin Gilbert, Steve Jacobs, Julie Firman
and Briana Sounhein provided important insight and data for the juvenile and adult coho salmon datasets collected by
the Oregon Department of Fish and Wildlife.
I would like to thank my friends for their tireless support through these years of graduate school. Dana Erickson,
Stephanie Gunckel, Peggy Kavanagh, Tara Nierenberg, Jennifer Gilden, Kristi Murphey, Mindi Sheer, and Michelle
Donaghy-Cannon have always been there to provide support, encouragement and the occasional tea break.
Finally, I would like to thank my family. My parents Pat and Jake Flitcroft inspired in me a love of nature from
my childhood. They have unconditionally loved and supported me in all that I have attempted. I would also like to
acknowledge my late brother Patrick, whose childhood curiosity about nature will forever inspire me. My husband Jeff
Snyder has always believed in me and encouraged me to learn and reach and grow. And my daughter Gwen consistently
reminds me of what’s most important in life.
CONTRIBUTION OF AUTHORS
Dr. Gordon H. Reeves was an advisor and editor on all manuscripts included in this work. Dr. Robert E. Gresswell
was an advisor and editor on Chapter 2. Dr. Lisa Ganio was an advisor, editor, statistical consultant, and statistical
programmer on Chapter 3. Dr. Alix Gitelman was an advisor, editor, and statistical consultant on Chapter 4. Kelly
Christiansen was a GIS analyst on Chapter 4.
TABLE OF CONTENTS
Chapter 1: Stream Networks and Multiple Spatial Scales for an Analysis of
Coho Salmon (Oncorhynchus kisutch) on the Oregon Coast
1
Introduction
1
Coho Salmon Life History
Study Goals
Network Framework: Chapter 2
Stream Networks and Scales for Juvenile Coho Salmon: Chapter 3
Hierarchical Bayesian Modeling: Chapter 4
Literature Cited
1
2
2
3
3
4
Chapter 2: Dynamic Network Topology: Towards Developing a Stream Network
Perspective
Abstract
Introduction
6
Defining Dynamic Network Topology
Life History Diversity and Dynamic Network Topology
Networks and Spatial Scale
Isolation and Movement in Stream Networks
Analysis and Dynamic Network Topology
An Application of Dynamic Network Topology
8
10
10
11
12
13
Conclusions
Literature Cited
14
14
Chapter 3: Exploring Network Patterns in the Habitat, Abundance and Distribution
of Juvenile Coho Salmon (Oncorhynchus kisutch) in Oregon’s Mid-Coast
Abstract
Introduction
Materials and Methods
Study Area
Datasets
Generating Stream Network Variables
Variable Selection
Objective One: Juvenile Density at the Site Scale
Objective Two: Juvenile Coho Patches
Objective Three: Patterns of Juvenile Occupancy in Subbasins
6
6
17
17
17
18
18
19
20
22
22
23
23
TABLE OF CONTENTS, continued
Chapter 3: Exploring Network Patterns in the Habitat, Abundance and Distribution
of Juvenile Coho Salmon (Oncorhynchus kisutch) in Oregon’s Mid-Coast (continued)
Results
Variable Selection
Objective One: Site Scale Pattern
Objective Two: Patches and Basin Shape
Objective Three: Subbasins and Juvenile Network Occupancy
24
24
24
28
32
Discussion
Objective One: Site Scale
Objective Two: Patch Scale
Objective Three: Subbasin Scale
Synthesis Among Scales
36
36
36
37
38
39
39
Conclusions
Literature Cited
Chapter 4: Multiple Spatial Scales in an Analysis of Two Life History Stages of Coho Salmon
(Oncorhynchus kisutch): A Hierarchical Bayesian Approach
55
55
55
Abstract
Introduction
Methods
Study Area
Datasets and Variables
Bayesian Methods and Models
58
58
58
60
Results
Adults
Juveniles
65
65
66
66
69
69
Discussion
Conclusion
Literature Cited
Chapter 5: How Important are Network Considerations Anyway?
72
72
73
Summary and Conclusions
Literature Cited
Bibliography
A list of appendices follows the list of figures and tables.
74
FIGURES
1.1. Outline of streams (5th order and larger) in the CLAMS study area.
2
2.1.
Historically, stream studies have focused on stream reaches or habitat units (a) rather than
analysis of the stream network as a whole (b).
7
2.2. As disturbance events contribute material to streams, and as stream hydraulics reorganize
substrates, the juxtaposition of stream habitats will change over time. For example, the
configuration in (a) may change to configuration (b) and then to configuration (c).
8
2.3
Direction and options for movement for aquatic obligate species such as fish are restricted
to upstream or downstream in a stream segment or laterally into tributaries compared to
species in a terrestrial system with potential routes on a 360° radius. The limited directions of
movement are a fundamental organizing feature of stream networks for movement of aquatic
species.
9
2.4
Network distances may reflect connectivity, but there is more to the orientation of stream
segments than simply the distance between them. Sites may be close to one another using a
Euclidean or “as the bird flies” distance measure, yet be far from one another using a network
measurement (as in sites 2 to 5). Also, sites that are far from one another using a network
distance (as in sites 2 to 5) may be more physically similar than sites that are close together
(as in sites 2 to 4) because of shared underlying landscape conditions (such as vegetation,
geology, elevation, and gradient).
9
2.5
Two extreme examples of habitat arrangements may be conceptually useful when considering
how an individual juvenile coho salmon must move within the network to find necessary
seasonal. Figures (a) and (b) share the same stream network, and the same numbers of habitat
sites classified for coho salmon as spawning (S), summer refuge (sum), or winter rearing
(win). However, the arrangement of the habitats is quite different with figure (a) depicted
interspersion of habitats while each seasonal habitat is restricted to specific tributaries in (b).
13
Highlighted subbasins in the Alsea and Siletz River Basins that were included in an analysis
of multiple spatial scales that incorporated a stream network framework.
19
3.2
The three spatial scales included in a multiple scale analysis of subbasins in the Alsea and
Siletz River basins are the site, patch and subbasin.
20
3.3
Generalized pictorial of network distance calculation: network distances were calculated
from juvenile snorkel sites to the closest available seasonal habitat.
22
Principle Components Analysis graph of 2002 habitat dataset displaying the number of
pieces of wood with subbasins of increasing or decreasing juvenile density (between 2001
and 2002) as an overlay. Each triangle corresponds with a specific pool and the size of the
triangle represents the relative number of pieces of wood associated with each pool. The
x-axis is associated a gradient of greater to smaller percentages of fine substrate material and
the y-axis is associated with smaller to greater percentages of coarse substrate material.
24
3.1
3.4
FIGURES
3.5
3.6
3.7
3.8
3.9
3.10
4.1
4.2
4.3
4.4
Principle Components Analysis graph of 2002 network dataset displaying distance to
spawning habitat with basins of increasing or decreasing juvenile density (between 2001
and 2002) as an overlay. Each triangle corresponds with a specific pool and the size of the triangle represents the measured distance to spawning habitat associated with each pool. The
x-axis is associated with a gradient of smaller to larger distances between seasonal habitats
while the y-axis represents a continuum of mean gradient downstream.
28
Juvenile coho salmon patch size (estimated variogram range) for 1998, 1999, 2001 and 2002
in subbasins of the Alsea and Siletz River basins. The subbasins are listed from smallest to
largest.
28
Variograms for subbasins in Oregon’s Mid-Coast (in each year available between 1998 and
2002) with arrows pointing to “hump” shape that is common in larger basins and may be
associated with a nested spatial autocorrelation structure.
29
Dendrogram showing two clusters of subbasins described as branched or linear in shape due
to their geomorphology.
32
The series of Canal Creek maps (1a, 1c, 1d) show an increase in juvenile network occupancy
upstream from 1998 to 2002 while Sunshine Creek (2a-2d) shows variation in network
occupancy including a decrease in occupancy between 1998 and 1999 and 2001 and 2002.
34
Number of coho salmon spawners in Oregon’s Mid-Coast and precipitation during the
spawning run (October through March).
35
Three spatial scales used in hierarchical Bayesian analysis of adult and juvenile coho salmon.
Basin and subbasin scales were used to hierarchically model adult coho salmon and subbasin
and site scales were used to hierarchically model juveniles.
57
Oregon’s coastal region encompassed the 5th-field and 4th-field hydrologic units used in a
multi-scale hierarchical Bayesian analysis.
57
Hierarchical Bayesian model structure for adult (a) and juvenile (b) coho salmon.
61
The caterpillar plot for juvenile coho salmon model #9c displays the posterior estimate and
95% interval for each parameter in the model. These intervals may be interpreted by examining
their overlap and whether they include zero. Intervals that include zero are not considered
to significantly contribute to the model result. Intervals that do not overlap display the
importance of modeling including the hierarchical scale of division. This allows variability
to be different with different parameters rather than needing to be shared throughout the
dataset.
67
TABLES
3.1
Variables in network and habitat datasets with source information and
transformations.
21
3.2
Summary of correlation matrix showing significant correlation relationships (> 0.60) used
in variable reduction from Oregon’s Mid-Coast region in 1999 (n = 987), 2001 (n = 1288),
and 2002 (n = 1416): Habitat and network datasets.
24
Principle components analysis results for sites in Oregon’s Mid-Coast in 1998, 1999, 2001
and 2002 using a dataset of variables describing network relationships.
25
Principle components analysis results for sites in Oregon’s Mid-Coast in 1998, 1999, 2001
and 2002 using a dataset of variables describing in-stream habitat conditions.
25
ANOVA results examining juvenile coho density and subbasins in Oregon’s Mid-Coast
region over time.
25
Pairwise comparisons of the difference in juvenile coho density between years for each
subbasin.
26
Local-mean non-parametric multiplicative regression model results for 1998, 1999, 2001
and 2002 for sites in Oregon’s Mid-Coast region. The dependent variable is log juvenile
coho density.
27
Discriminant Analysis results for two sets of analysis of groups in subbasins of Oregon’s
Mid-Coast: groups of subbasins identified as increasing or decreasing juvenile coho density
(between 2001 and 2002); groups of subbasins classified as linear or branched in shape.
32
Summary table showing number of sites occupied, total number of sites surveyed, juvenile
coho count, number of streams occupied and stream miles occupied for the 11 subbasins of
interest in Oregon’s Mid-Coast region.
33
Means and confidence intervals for annual estimates of the number of stream kilometers
occupied by juvenile coho salmon between 1998 and 2002 averaged over all subbasins.
33
Pairwise comparisons among years of the number of stream kilometers occupied by juvenile
coho salmon between 1998 and 2002. T-statistic is testing for no difference between years.
33
Variables that represent different spatial scale were used in multi-scale hierarchical Bayesian
models of the adult and juvenile life history stages of coho salmon. *Variable was standardized.
59
Datasets that were assembled to derive variables that could be used to represent multiple
spatial scales.
59
The number of 5th field HUCs within 4th field HUCs and the number of sites sampled in
each 5th field HUC.
60
3.3
3.4
3.5
3.6
3.7
3.8
3.9
3.10
3.11
4.1
4.2
4.3
TABLES
4.4
4.5
4.6
4.7
4.8
Bayesian hierarchical models for adult coho salmon on the Oregon coast. For each model,
subbasin-scale variables that are hierarchically connected to basin-scale variables are listed
next toone another in the table columns. The Bayesian Information Criterion or BIC value
allows for a comparison of model effectiveness by capturing the residual variance and a
penalty term. The model with the lowest BIC value is considered to be the best fitting model.
62
Bayesian hierarchical models for juvenile coho salmon in subbasins of Oregon’s mid-coast.
For each model, site-scale variables that are hierarchically connected to subbasin-scale
variables and are listed next to one another in the table columns. The Bayesian Information
Criterion or BIC value allows for a comparison of model effectiveness by capturing the
residual variance and a penalty term. The model with the lowest BIC value is considered to
be the best fitting model.
64
Bayesian Information Criterion (BIC) results for adult coho salmon models from 2001 and
2002. BIC allows for a comparison of model effectiveness by capturing the residual variance
and a penalty term. The model with the lowest BIC value is considered to be the best fitting
model. Models 5b and 14b were created by altering the set of variables in models 5 and 14
respectively.
65
Important combinations of variables from Bayesian hierarchical models of the mean of adult
coho salmon at a subbasin scale.
66
Bayesian Information Criterion (BIC) results for juvenile coho salmon models from 2002.
BIC allows for a comparison of model effectiveness by capturing the residual variance and a
penalty term. The model with the lowest BIC value is considered to be the best fitting model.
Models 9b and 9c were created by altering variables in model 9.
66
APPENDICES
Appendix 3.1: ODFW Aquatic Inventory and Analysis Project: Habitat Benchmarks
42
Appendix 3.2: Variograms for small basins in Oregon’s Mid-Coast region in each year available
between 1998 and 2002.
43
43
Figure Series 3.2.1-3.2.5
Appendix Figure 3.2.1. Canal Creek Variograms from 1998, 2001 and 2002.
Appendix Figure 3.2.2. Cedar Creek Variograms from 1999, 2001 and 2002.
Appendix Figure 3.2.3. Sams Creek Variograms from 1999, 2001 and 2002.
Appendix Figure 3.3.4. Fall Creek Variograms from 1998, 2001 and 2002.
Appendix Figure 3.3.5. Sunshine Creek Variograms from 1998, 1999, 2001 and 2002.
Appendix 3.3
43
43
44
44
45
46
Rebecca L. Flitcroft. 2007. Regions to streams: Spatial and temporal Variation in stream occupancy patterns of coho salmon (Oncorhynchus
kisutch) on the Oregon Coast. Ph.D. dissertation. Department of Fisheries and Wildlife, Oregon State University, Corvallis, OR.
Chapter 1
Stream Networks and Multiple Spatial Scales for an Analysis of Coho Salmon
(Oncorhynchus kisutch) on the Oregon Coast
Introduction
Research that meaningfully contributes to understanding
patterns of occupancy for an aquatic obligate species using
multiple spatial scales and a network perspective is unusual and
may expand how ecologists think about aquatic organization.
This dissertation focuses on juvenile and adult coho salmon
(Oncorhynchus kisutch) in coastal Oregon streams. Patterns
of occupancy by juvenile coho salmon populations in stream
networks of Oregon’s mid coast are explored, as are adult
coho salmon populations between Astoria in the north and
Cape Blanco to the south (Figure 1.1). Not only do these
boundaries correspond with the area studied by the Coastal
Landscape Analysis and Modeling Study (CLAMS) from
which many key datasets are derived, but they correspond
with coastal ecoregions that describe areas of relatively
similar geology, vegetation and precipitation (Thorson et al.
2003). Coho salmon were selected for analysis in this study
primarily because of their over-winter use of coastal streams,
the relatively small stream size (1st–3rd-order on 1:100,000
scale maps; Strahler 1952) occupied by juveniles, and the
wide distribution of migrating adults. Also, abundant data are
available that document juvenile coho salmon distribution,
abundance and habitat as well as adult spawning numbers and
locations.
Perhaps the most important criteria for selecting coho
salmon in an analysis of stream networks is that coho salmon
life history, like the histories of other salmonids, is a complex
example of adaptation and survival that is dependent on a
diverse array of habitats. Adult and juvenile coho salmon use
the stream network in different ways. Adults migrate through
the stream system in search of spawning habitat and partners.
Juveniles navigate the stream network in search of adequate
seasonal habitats for a year of freshwater residency before
migrating to the ocean. Each life history stage includes a
variety of behaviors that are adapted to existence in a changing
stream environment. I will explore the importance of habitat
diversity over time for coho salmon with a network perspective
that uses the complexity of a stream network for insight and
investigation.
Coho Salmon Life History
Coho salmon, also known as silver salmon, range in North
America between Monterey Bay, California and Point Hope,
Alaska. Coho salmon are one of five species of Pacific salmon
belonging to the genus Oncorhynchus. The other four species
of Pacific salmon are pink (O. gorbuscha), chum (O. keta),
sockeye (O. nerka) and Chinook (O. tshawytscha) salmon.
Chinook and coho salmon are found in Oregon. The two
species of anadromous trout that are also found in Oregon are
steelhead (O. mykiss) and coastal cutthroat (O. clarkii). (Groot
and Margolis 1991, Lichatowich 1999, Quinn 2005).
Coho salmon spawning run times vary by latitude, with
adults entering rivers in Alaska in July, British Columbia
in September and October and Oregon and California in
November and December. However, some specific streams
have migration runs of coho salmon as early as April or as late
as February and March (Groot and Margolis 1991). Juvenile
coho salmon in Oregon and Washington typically spend their
first year in smaller streams and migrate to the ocean in the
spring of their second year (Lichatowich 1999). Habitat needs
for juvenile coho salmon differ between summer and winter
(Nickelson et al. 1992a, Sleeper 1993, Spalding et al. 1995).
Deep pools and cool water characterize summer habitat, and
winter habitat is characterized by large wood and off-channel
refuges such as backwaters, alcoves, secondary channels or
tributaries (Quinn and Peterson 1996, Rosenfeld et al. 2000,
Bell 2001, Bramblett et al. 2002).
The majority of adult coho salmon return to their natal
streams to spawn. Unfortunately, research on salmon straying
rates is limited. Labelle (1992) reported that approximately
4.7% of coho salmon strayed between nine streams on the coast
of Vancouver Island. However, he conjectured that straying
rates could be more than 40% in some streams for some years.
Straying in other salmon species such as Chinook, sockeye
and pink salmon and steelhead has also been documented
(Shapovalov and Taft 1954, Cooper and Mangel 1998).
Quinn (2005) hypothesized that straying would be more
common in species that experience greater habitat disturbance
Flitcroft
Chapter 1
2
Figure 1.1. Outline of streams (5th order and larger) in the CLAMS study area.
than in those with more predictable environments. In this
case, straying would be an adaptation to ensure the survival of
the population in inconsistent environments. This argument
and the literature that documents the common incidence of
straying support the assertion that metapopulation structure is
possible and even likely in some populations of salmon.
Study Goals
This study seeks to accomplish three goals:
1) Develop a comprehensive framework in which to consider
spatial patterns of aquatic obligate species that incorporates
the physical context of a stream network.
2) Explore juvenile coho distribution at multiple spatial scales
including an assessment of the usefulness of stream network
orientations of habitats or spatial position in understanding
juvenile stream use.
3) Incorporate multiple spatial scales into a hierarchical
Bayesian model for adult and juvenile coho salmon
densities.
Network Framework: Chapter 2
I recognized the usefulness of an articulated network
framework for ecological research once work on this
dissertation began in earnest. I was challenged by the paucity
of ecological theory and analytical tools that were focused
on stream networks. While a network perspective has been
explored by some researchers, much work is left to be done. For
example, Gresswell et al. (2006) explored network patterns for
assessment of cutthroat trout distributions. Isaak and Thurow
(2006) explored patterns of Chinook redd distribution in
a central Idaho basin. They found that use of the stream
network increased with increasing population growth. Smith
Flitcroft
Chapter 1
and Kraft (2005) incorporated network position variables
when evaluating fish assemblages in a watershed in New
York. They found that network variables, such as confluence
link, provided an avenue to greater ecological understanding
of the metrics that describe fish distribution. Kocik and
Ferreri (1998) explored the juxtaposition of habitats and size
of habitats when assessing juvenile Atlantic salmon (Salmo
salar) growth and survival. Their model of habitat showed
that understanding habitat placement and size was relevant
when considering juvenile survivorship. Beyond these
limited studies, little has been done to incorporate network
relationships in either ecological understanding or analytical
tools for exploring aquatic patterns of life in stream networks.
Currently, network-based calculations of habitat relationships
are absent from most current habitat interpretation or analysis.
In order to help facilitate further ecological research
in streams that include a network perspective, I offer the
dynamic network topology framework. This framework is
meant to include the dynamic nature of stream habitats over
time with the complexity of travel options unique to stream
networks due to limited available directions of movement and
the inherent flow of water, nutrients and inorganic matter.
Stream Networks and Scales for Juvenile Coho
Salmon: Chapter 3
The context of the stream network will be explored in an
attempt to further our understanding of distribution patterns
of juvenile and adult coho salmon. Patterns within stream
networks were identified by Gresswell et al. (2006) as an
important research direction for analysis of biotic and abiotic
relationships. Network distance measurements and variables
that describe network position will be incorporated into the
analysis of juvenile patterns of distribution. This analysis
included multiple spatial scales in an attempt to explore
interactions between ecological processes and fish distribution
in the network. Three spatial scales will be considered: site, patch
and subbasin. The site scale corresponds to local conditions
and the juxtaposition of habitats within the network, and
site level conditions such as substrate will be considered in an
exploration of juvenile density. The patch scale encompasses
the area of spatial autocorrelation of juvenile coho salmon
within the stream network, and will provide a comparison of
patterns of juvenile occupancy across subbasins. The subbasin
scale allows for a comparison of larger patterns of juvenile
occupancy with large scale environmental conditions such as
precipitation and adult spawning run size.
Spatial location of stream features has been explored in
assessments of fish assemblage (Megalhaes et al. 2002, Benda
et al. 2004) or biotic diversity (Day et al. 1992, Wright and
Li 2002) as researchers attempt to link patterns with biotic
distributions. Burnett (2001) explored the use of different
spatial extents over time in describing the distribution
3
of juvenile salmonid assemblages in the Elk River Basin,
Oregon. She found that both spatial and temporal extent were
important in describing the observed variability in assemblage
distribution. Geomorphic criteria have been used extensively
to define stream reaches (Hankin and Reeves 1988, Moore
et al. 1997, Rosgen 1994). Because fish habitat is associated
with stream geomorphic characteristics, physical descriptions
of streams have been helpful in characterizing seasonal
habitat availability (Nickelson 1998, Jones and Moore 1999,
Rosenfeld et al. 2000) and for use in predictions of population
growth and survival (Nickelson et al. 1992b, Lichatowich et
al. 1995, Quinn and Peterson 1996). Landscape features have
also been used to predict fish distribution and abundance
(Bradford et al. 1997, Lunetta et al. 1997, Argent et al. 2003).
Employing a network perspective when considering
juvenile coho salmon abundance may allow for novel
ecological interpretations of occupancy. For example, work
by Chapman (1962) that identified two types of juvenile
behavior was reinterpreted decades later from a different
perspective (Nielsen 1992). Chapman identified territorial
coho salmon and “nomads” that moved around and did not
compete successfully with the territorial fish. He concluded
that the “nomads” likely left the stream system and perished.
Similar studies of behavior by Nielsen (1992) indicated that
the non-territorial coho salmon may indeed not grow as
quickly as territorial fish, but that the choice to move could be
a successful life history strategy for these fish. Such a shift in
interpretation of empirical observations may be possible when
considering behavior and abundance within the context of the
stream network.
Hierarchical Bayesian Modeling: Chapter 4
I reasoned that different spatial scales may correspond to
the needs, drives, and limitations of fish at different points
in their life history. For example, a large spatial scale might
correspond with the dispersal of adult fish while a small spatial
scale might correspond with movement patterns of small fish.
I also reason that differences in spatial scales may contribute
to a greater understanding of distribution patterns when
hierarchically nested in an analysis that considers life history
stage along with habitat needs.
Three spatial scales, site, subbasin, and basin, were used
in this analysis of coho salmon. Adult coho salmon were
hierarchically modeled using the subbasin and basin scales and
juvenile coho were modeled with the site and subbasin scales.
Two years of data were used to model adults (2001 and 2002)
with only one year (2002) modeled for juvenile fish. Larger
scale variables were used with adult salmon to reflect their
movement potential and their spawning drive. Smaller scales
were used for juveniles to reflect their limited mobility and
their need to move among habitats to survive in freshwater for
a year before migrating to the ocean.
Flitcroft
Chapter 1
Literature Cited
Argent, D.G., J.A. Bishop, J.R. Stauffer, Jr., R.F. Carline and W.L.
Myers. 2003. Predicting freshwater fish distributions using
landscape-level variables. Fisheries Research 60: 17-32.
Bell, E. 2001. Survival, growth and movement of juvenile coho
salmon (Oncorhynchus kisutch) over-wintering in alcoves,
backwaters and main channel pools in Prairie Creek, California.
Masters thesis. Humboldt State University, Arcata, California.
Benda, L., L. Poff, D. Miller, T. Dunne, G. Reeves, G. Pess and M.
Pollock. 2004. The network dynamics hypothesis: how channel
networks structure riverine habitats. BioScience 54(5): 413-427.
Bradford, M.J., G.C. Taylor and J.A. Allan. 1997. Empirical review
of coho salmon smolt abundance and the prediction of smolt
production at the regional level. Transactions of the American
Fisheries Society 126: 49-64.
Bramblett, R.G., M.D. Bryant, B.E. Wright and R.G. White. 2002.
Seasonal use of small tributary and main-stem habitats by juvenile
steelhead, coho salmon, and dolly varden in a Southeastern Alaska
drainage basin. Transactions of the American Fisheries Society
131: 498-506.
4
Jones, K.K. and K.M.S. Moore. 1999. Habitat assessment in coastal
basins in Oregon: implications for coho salmon production and
habitat restoration. Pages 329-340 in E.E. Knudsen et al. Editors.
Sustainable Fisheries Management. CRC Press, New York.
Kocik, J.F. and C.P. Ferreri 1998. Juvenile production variation in
salmonids: population dynamics, habitat, and the role of spatial
relationships. Canadian Journal of Fisheries and Aquatic Sciences
55(Suppl. 1): 191-200.
Labelle, M. 1992. Straying patterns of coho salmon (Oncorhynchus
kisutch) stocks from southeast Vancouver Island, British Columbia.
Canadian Journal of Fisheries and Aquatic Science 49: 1843-1855.
Lichatowich, J. 1999. Salmon without rivers a history of the Pacific
salmon crisis. Island Press, Coveto, California.
Lichatowich, J., L. Mobrand, L Lestelle and T. Vogel. 1995. An
approach to the diagnosis and treatment of depleted Pacific
salmon populations in freshwater ecosystems. Fisheries (Bethesda)
20(l): 10-18.
Lunetta, R.S., B.L. Cosentino, D.R. Montgomery, E,M. Beamer
and T.J. Beechie. 1997. GIS-based evaluation of salmon habitat in
the Pacific Northwest. Photogrammetric Engineering & Remote
Sensing 63(10): 1219-1229.
Burnett, K.M. 2001. Relationships among juvenile anadromous
salmonids, their freshwater habitat, and landscape characteristics
over multiple years and spatial scales in the Elk River, Oregon.
Doctoral dissertation. Oregon State University, Corvallis, Oregon.
Megalhaes, F.M., D.C. Batalha and M.J. Collares-Pereira. 2002.
Gradients in stream fish assemblages across a Mediterranean
landscape: Contributions of environmental factors and spatial
structure. Freshwater Biology 47: 1015-1031.
Chapman, D.W. 1962. Food and space as regulators of salmonid
populations in streams. The American Naturalist 100(913): 345357.
Moore, K.M.S., K.K. Jones and J.M. Dambacher. 1997. Methods
for stream habitat surveys. Oregon Department of Fish and
Wildlife, Information Report 97-4, Portland, Oregon.
Cooper, A.B. and M. Mangel. 1998. The dangers of ignoring
metapopulation structure for the conservation of salmonids.
Fisheries Bulletin 97: 213-226.
Nickelson, T.E. 1998. A habitat-based assessment of coho salmon
production potential and spawner escapement needs for Oregon
coastal streams. Oregon Department of Fish and Wildlife
Information Report No. 98-4, Corvallis, Oregon.
Day, D.M., W.A. Bertrand, M.J. Wiley and R. Sauer. 1992.
Influence of stream location in a drainage network on the index
of biotic integrity. Transactions of the American Fisheries Society
121: 635-643.
Gresswell, R.E., C.E. Torgersen and D.S. Bateman. 2006. A
spatially explicit approach for evaluating relationships among
coastal cutthroat trout, habitat, and disturbance in headwater
streams. Pages 457-471 in R. M. Hughes, L. Wang and P. W.
Seelbach, editors. Influences of landscapes on stream habitats and
biological assemblages. American Fisheries Society, Symposium
48, Bethesda, Maryland.
Groot, C. and L. Margolis, editors. 1991. Pacific Salmon Life
Histories. University of British Columbia Press, Vancouver, B.C.,
Canada.
Hankin, D.G. and G.H. Reeves. 1988. Estimating total fish
abundance and total habitat area in small streams based on visual
estimation methods. Canadian Journal of Fisheries and Aquatic
Sciences 45: 834-844.
Isaak, D.J. and R.F. Thurow. 2006. Network-scale spatial and
temporal variation in Chinook salmon (Oncorhynchus tshawytscha)
redd distributions: patterns inferred from spatially continuous
replicate surveys. Canadian Journal of Fisheries and Aquatic
Sciences 63(2): 285-296.
Nickelson, T.E., J.D. Rodgers, S.L. Johnson and M.F. Solazzi.
1992a. Seasonal changes in habitat use by juvenile coho salmon
Oncorhynchus kisutch in Oregon coastal streams. Canadian
Journal of Fisheries and Aquatic Sciences 49(4): 783-789.
Nickelson, T.E., M.F. Solazzi, S.L. Johnson and J.D. Rodgers.
1992b. An approach to determining stream carrying capacity and
limiting habitat for coho salmon Oncorhynchus kisutch. Pages 251260 in L. Berg and P.W. Delaney, editors. Proceedings of the coho
workshop. Naniamo, B.C., Canada.
Nielsen, J. 1992. Microhabitat-specific foraging behavior, diet, and
growth of juvenile coho salmon. Transactions of the American
Fisheries Society 121: 617-634.
Quinn, T.P. 2005. The behavior and ecology of Pacific salmon and
trout. University of Washington Press, Seattle, Washington.
Quinn, T.P. and N.P. Peterson. 1996. The influence of habitat
complexity and fish size on over-winter survival and growth of
individually marked juvenile coho salmon (Oncorhynchus kisutch)
in Big Beef Creek, Washington. Canadian Journal of Fisheries and
Aquatic Sciences 53: 1555-1564.
Flitcroft
Chapter 1
Rosenfeld, J., M. Porter and E. Parkinson. 2000. Habitat factors
affecting the abundance and distribution of juvenile cutthroat
trout (Oncorhynchus clarki) and coho salmon (Oncorhynchus
kisutch). Canadian Journal of Fisheries and Aquatic Sciences 57:
766-774.
Rosgen, D.L. 1994. A classification of natural rivers. Catena 22:
169-199.
Shapovalov, L. and A.C. Taft. 1954. The life histories of the steelhead
rainbow trout (Salmo gairdneri gairdneri) and silver salmon
(Oncorhynchus kisutch) with special reference to Wadel Creek,
California, and recommendations regarding their management.
Fish Bulletin of the California Department of Fish and Game
98:1-375.
Sleeper, Jack. 1993. Seasonal changes in distribution and abundance
of salmonids and habitat availability in a coastal Oregon basin.
Masters thesis. Oregon State University, Corvallis, Oregon.
Smith, T.A. and C.E. Kraft. 2005.Stream fish assemblages in relation
to landscape position and local habitat variables. Transactions of
the American Fisheries Society 134: 430-440.
5
Spalding, S., N.P. Peterson and T.P. Quinn. 1995. Summer
distribution, survival, and growth of juvenile coho salmon
under varying experimental conditions of brushy instream cover.
Transactions of the American Fisheries Society 124: 124-130.
Strahler, A.N. 1952. Dynamic basis of geomorphology. Geological
Society of America Bulletin 63: 923-938.
Thorson, T.D., S.A. Bryce, D.A. Lammers, A.J. Woods, J.M.
Omernik, J. Kagan, D.E. Pater and J.A. Cornstock. 2003.
Ecoregions of Oregon (Color poster with map, descriptive test,
summary tables, and photographs). U.S. Geological Survey (map
scale 1:1,500,000). Reston, Virginia.
Wright, K.K. and J.L. Li. 2002. From continua to patches:
examining stream community structure over large environmental
gradients. Canadian Journal of Fisheries and Aquatic Sciences 59:
1404-1417.
Flitcroft
Chapter 2
6
Rebecca L. Flitcroft. 2007. Regions to streams: Spatial and temporal Variation in stream occupancy patterns of coho salmon (Oncorhynchus
kisutch) on the Oregon Coast. Ph.D. dissertation. Department of Fisheries and Wildlife, Oregon State University, Corvallis, OR.
CHAPTER 2
Dynamic Network Topology:
Towards Developing a Stream Network Perspective
Rebecca L. Flitcroft
Department of Fisheries and Wildlife
Oregon State University
Corvallis, Oregon
Abstract
As the broader discipline of ecology embraces dynamic processes as a fundamental element of ecosystems, aquatic
ecologists are working to develop theory and techniques that allow for the analysis of dynamic stream processes and
communities of organisms. The stream network is the basis upon which stream habitats are organized. Natural and
anthropogenic disturbances in streams alter the configuration of stream habitats seasonally, decadally, or over centuries.
Native aquatic species have developed different mechanisms for coping with the dynamic nature of habitats in stream
networks. However, stream network structure complicates habitat connectivity for aquatic obligate species due to limited
directions of movement offered by stream channels and the inherent directionality of flow of water and other elements of
the stream environment. A conceptual framework that includes the dynamic nature of streams within the context of the
stream network may by useful for the study and conservation of aquatic species. Dynamic network topology is proposed
as a conceptual framework that includes the location and juxtaposition of habitats in stream systems with the underlying
structure of the stream network. Pacific salmon and trout are used to discuss the implications of dynamic network
topology for aquatic obligate species. Ultimately, a perspective that includes the constraints and opportunities for stream
species of network configurations may help in the development of analytical tools and management recommendations
that benefit native species. Aquatic obligate species that reside in streams offer ecologists an opportunity to explore new
ways of thinking about movement and behavior by using the stream network as a critical organizing feature describing
biotic diversity.
Introduction
Ecology is an evolving science. A guiding principle in
ecology has been the concept of a steady state (Wallington
2005), or climax community reached through succession
(Stiling 1992). Disturbance has been viewed as a damaging
disruption of systems headed for stability (Horn 1981).
While the concept of instability as the normal working
arrangement of ecosystems is not new (Gleason 1926, 1927),
it has only been relatively recently that ecologists have begun
to embrace the dynamic spatial and temporal patterns of
habitats, disturbance, and organisms that are now considered
a fundamental aspect of the environment (Naiman et al.
1992). This shift in philosophy has repercussions throughout
ecological theory and investigation, and has been considered
by some to be a paradigm change (Wallington 2005).
Several concepts and approaches to aquatic ecology are
consistent with the static environment paradigm. For example,
stream fishes were often investigated from the perspective of
the restricted-movement paradigm (Gerking 1959), which
held that adult resident fish were generally sedentary and
spent the majority of their lives within short stream reaches
(20-50 m). Lack of movement by resident fishes is consistent
with an ecological perspective that views streams as static and
unchanging. Consistent with this philosophy is the common
aquatic methodological approach of sampling individual sites
within a stream system, rather than entire streams (Richards
and Cernera 1989, Silliman and Bertness 2002) (Figure 2.1).
This sampling method assumes that fish are not using other
areas thereby making sampled sites “representative” of other
unsampled sites, or that movement is either uncommon, or
maladaptive (Chapman 1962).
Flitcroft
Chapter 2
Figure 2.1. Historically, stream studies have focused on stream
reaches or habitat units (a) rather than analysis of the stream
network as a whole (b).
In recent decades, aquatic ecologists have followed the
general trend in ecology and have begun the shift toward a
worldview that considers dynamic natural processes as integral
and necessary to maintaining ecological diversity (Naiman et
al. 1992). The linear (Vannote et al. 1980) or static (Frissell
et al. 1986) descriptions of stream organization that have
dominated aquatic investigation (Fisher 1997) do not
adequately represent a philosophical perspective that focuses
on interconnected and dynamic processes. More recent
organizational frameworks consider entire river systems, or
“riverscapes” (Fausch et al. 2002), and include disturbance as a
dynamic and integrated aspect of stream communities (Benda
et al. 2004).
Aquatic ecologists have begun to recognize the importance
of expanding the spatiotemporal scale of consideration,
particularly as focus has shifted from individuals to the
restoration and preservation of entire populations, species,
and ecosystems (i.e., Scott et al. 1987, Li et al. 1995, Thurow
et al. 1997). By expanding the scale of analysis from sites
and individuals to watersheds and populations, patterns of
organisms or physical processes may be interpreted differently.
For example, it is possible that the decline of anadromous
Pacific salmon (Oncorhynchus spp.) in Oregon was not
identified for several decades because spawning surveys were
conducted only on the same prime spawning sites every
year (Cooney and Jacobs 1995). By looking at patches of
individuals in isolation from the entire population, changes
in habitat and salmon abundance at larger scales were missed.
While this may also be a sampling design issue, it shows that
population assessments are most accurate when they include
7
the range of conditions in which individuals of the population
must survive.
As aquatic ecologists begin to consider entire systems and
populations, the network architecture of streams and the
effects of network patterns of habitats on the distribution,
abundance, and movement of aquatic obligate species is a
conceptual challenge. Stream channels are the foundation
upon which habitats for aquatic species are organized, and
are configured as a network of branching, interconnected
tributaries. The organization of habitats within the network
changes in response to a variety of environmental conditions,
including disturbance (Benda et al. 1998, Nakamura et al.
2000) and discharge (Gardner et al. 2003, Poff et al. 1997).
For example, during low flows, fewer stream channels will
have water than at high flows. At high flows, there is generally
greater connectivity within the network as water fills dry
channels. With changing water levels, habitats in streams also
change. Summer pools flood and become winter riffles, and
dry channels fill with water becoming pools. These changes
in water levels, network connectivity, and habitat types mean
that the configuration of habitats for aquatic organisms also
changes. Therefore, individuals must be adapted to the new
habitats, or be able to move to other areas of the network. The
arrangement of habitats and changes in their configuration
may be an important consideration of population viability
assessments of aquatic species as individuals move to find
adequate habitats for survival.
Scientists have observed that stream networks are in
a constant state of flux in response to changing climate
conditions (seasonally, decadally, or over centuries), natural
disturbances (e.g., fires and landslides) and human activities
(Reeves et al. 1995). Benda et al. (2004) proposed the network
dynamics hypothesis that combines the hierarchical physical
structure of stream networks with dynamic disturbance
processes. By combining structure and process, changes in
fluvial geomorphology due to disturbances such as debris
flows may be predicted. For example, Benda et al. hypothesize
that tributary junctions will be a nexus for habitat diversity,
because it is at junctions that material from debris flows and
floods will be physically forced to halt. Debris flow material
contributes the raw materials necessary for habitat creation
(Benda et al. 1998, Landres et al. 1999, Reeves et al. 2002).
Patterns of deposition may be important biologically.
Torgersen et al. (2004) found that periodicity of freshwater
distribution patterns of coastal cutthroat trout (O. clarki
clarki) along a mainstem stream were associated with tributary
junctions. Locally, disturbances change the proximity and
connection of stream habitats in the network over varying
temporal and spatial scales. Close proximity of habitats
necessary for different life history stages has been associated
with higher fish densities and abundance (Kocik and Ferreri
1998, Flitcroft 2007). At larger scales, disturbances isolate
some habitats and connect others. In coastal Oregon, Reeves
Flitcroft
Chapter 2
et al. (1995) described the complexity of stream systems over
time in response to large scale disturbances. They noted that
persistence in a dynamic environment required organisms
to develop behaviors (e.g., movement, life-history diversity,
aestivation) adapted to change (Warren and Liss 1980). The
authors posited that landscape management that mimicked
the extent and frequency of natural events would be less
harmful than anthropogenic disturbances for native species
that are adapted to natural disturbance regimes.
Ecologists have long observed that organisms are adapted
to live in their environments (Southwood 1977). It intuitively
follows that native aquatic species are adapted to survive in
the changing habitat configurations of stream networks. For
example, different behaviors among individuals will be more or
less adaptive for the population depending on stream network
conditions. For instance, Quinn (2005) hypothesized that
adult anadromous salmonids might stray (fail to home to natal
streams to spawn) more in years characterized by disturbance
of natal stream beds. Adults that stray may be more successful
than those that home in years when natal spawning beds have
been disturbed. Therefore, straying endures within populations
of salmonids. Other behaviors present in salmonid populations
that essentially spread out risk include: some precocious males
return early to spawn (jacks); females leave multiple pockets
of eggs in each redd (Quinn 2005); and when juvenile salmon
hatch, some will move, some will stay, some juveniles will be
territorial and others will be “floaters” (Nielsen 1992). The
diversity of behaviors exhibited by a population of individuals
reflects evolutionary adaptation to dynamic stream conditions
that unpredictably favor one set of behaviors over another.
Analysis of populations of aquatic organisms could benefit
by expanding the perspective of ecologists to include the
complexity of spatial arrangements of habitat and movement
pathways in stream networks over time. In this paper, I offer
a dynamic network topology as a conceptual framework
that encompasses the elements of network connectivity,
juxtaposition of habitats, disturbance, and directionality of
flow (for water, nutrients, and individual aquatic organisms)
that may be critical to understanding adaptations in life
history and behavior of native species. I will first describe
dynamic network topology. Then in successive sections of the
paper, I will discuss dynamic network topology with regard
to: differences in network use depending on life history stage,
the implications of spatial scale, isolation of habitats and
movement of individuals, analysis of stream networks, and a
hypothetical scenario that applies dynamic network topology.
In dynamic network topology, the stream network is
the template upon which habitats are arranged (topology).
Resilience reflects the ability of native species to respond to the
diversity and changing topology of stream network habitats
that are dynamic over time. In this paper, Pacific salmon
and trout (Oncorhynchus spp.) are generally used to explore
ideas about network relationships and long-term survival
8
because of the diversity of salmonid life history types (Pacific
Fisheries Resource Conservation Council 1998), their stream
occupancy, and the long history of research studying salmonid
behavior and habitat needs.
Defining Dynamic Network Topology
Network topology is a universal notion that can inform
assessments of population viability, habitat quality and
quantity, and the interrelationship between aquatic species
and the stream environment. Stream network topology
principally describes the orientation and connection of
habitats to one another in a stream network. However, stream
habitats change over multiple spatial extents and temporal
scales, and therefore, the topology of an entire stream network
is in flux. Dynamic network topology, therefore, describes the
orientation of habitats that change over time a stream network
(Figure 2.2). The language of dynamic network topology is
meant to capture the changing nature of habitats in a stream
system in analysis that maintains the spatial context of the
underlying physical network.
Figure 2.2. As disturbance events contribute material to streams,
and as stream hydraulics reorganize substrates, the juxtaposition
of stream habitats will change over time. For example, the
configuration in (a) may change to configuration (b) and then to
configuration (c).
Flitcroft
Chapter 2
9
Figure 2.3. Direction and
options for movement for
aquatic obligate species
such as fish are restricted to
upstream or downstream in
a stream segment or laterally
into tributaries compared to
species in a terrestrial system
with potential routes on a
360° radius. The limited
directions of movement are
a fundamental organizing
feature of stream networks
for movement of aquatic
species.
Containment of water in channels limits movement
directions of organisms and abiotic elements to upstream,
downstream or peripherally into the hyporheic zone (Vannote
et al.1980, Fausch et al. 2002). In terrestrial systems, a
habitat can connect directly to other habitats along routes
extending potentially 360 degrees. In a stream system, by
contrast, a habitat directly connects to other habitats along
only four routes at most (upstream or downstream in a main
channel, and upstream into at most two tributaries) (Figure
2.3). Further, this branching of a stream network at tributary
junctions means that although two stream habitats (such as a
pool and a riffle) may be near one another in two-dimensional
Cartesian space, they may be very far apart in network distance
space (Figure 2.4, sites 2 and 5) (Gresswell et al. 2006, Lowe
et al. 2006). Aquatic organisms have few directions in which
to move, and habitats may be far from one another in the
network. Therefore, it may be difficult for aquatic species to
navigate between habitats, or even find particular habitats in
the stream network. Difficulty navigating the network may
result in reduced survival of aquatic organisms. Analysis
of habitats that neglects to consider movement paths and
directions for aquatic species may misrepresent population
viability.
Another consideration of habitat juxtaposition in stream
networks is that habitats may be far from one another in
network space but have similar physical characteristics (e.g.,
depth, amounts of large wood, substrate composition) because
they are embedded in similar landscapes (e.g., rock type,
vegetation, disturbance history, or land use) (Figure 2.4, sites
1, 2 and 5). It is also possible that stream habitats with similar
physical characteristics such as substrate (Figure 2.4, sites 2
and 5) may be further from one another in network space than
they will be to very dissimilar stream segments (Figure 2.4, site
2 to 4; or site 5 to 6). Therefore, network proximity between
Figure 2.4. Network distances may reflect connectivity,
but there is more to the orientation of stream segments
than simply the distance between them. Sites may be
close to one another using a Euclidean or “as the bird
flies” distance measure, yet be far from one another using
a network measurement (as in sites 2 to 5). Also, sites
that are far from one another using a network distance
(as in sites 2 to 5) may be more physically similar than
sites that are close together (as in sites 2 to 4) because
of shared underlying landscape conditions (such as
vegetation, geology, elevation, and gradient).
Flitcroft
Chapter 2
habitats does not necessarily describe habitat similarity. The
juxtaposition of habitat is further complicated seasonally
as water level fluctuations increase or decrease the physical
connectivity among stream habitats. Therefore, the topology
of habitats will change as different parts of the network are
cut off or reconnected, depending on discharge (Heggenes
et al. 1991). Taken together, these considerations imply that
network distance can be used to orient habitats to one another,
but the landscape and seasonal context is also important.
Combining habitat juxtaposition with landscape conditions
may allow biologists to better understand why organisms
are distributed as they are. This in turn may allow for more
accurate population level predictions of persistence.
Depending upon the ecological process of interest, the
effective distance between two habitats in a stream network
may be greater in one direction than in another, reflecting
the influence of flow direction. The effect of flow will vary
with life history stage and mobility of a species. For example,
while two rearing habitats may be equally spaced upstream
or downstream from a spawning site, a newly emerged trout
alevin may be more able to move downstream with the flow of
the stream than upstream and against the flow (Ottaway and
Forrest 1983). As the alevin grows, movement upstream may
become less difficult (Kahler et al. 2001). Thus, the strength of
connection between habitats (i.e., permeability of boundaries)
can vary with distance and flow direction, and may most
productively be considered with respect to the mobility of
individual organisms. Habitat patch size, permeability of
habitat patch boundaries, and spatial isolation or connectedness
of patches may be considered from a network perspective by
using the framework of dynamic network topology. Dynamic
network topology means that the location of habitats relative to
one another is coupled with a riverscape context (i.e., geology,
geomorphology, land use, disturbance history, or elevation),
therefore providing an inclusive perspective for assessments of
populations of aquatic species.
Life History Diversity and Dynamic Network
Topology
Warren and Liss (1980) described capacity of a system as “all
possible developmental states and all possible performances
that a system may exhibit while still maintaining its integrity
as a coherent entity.” The capacity for species to survive in a
dynamic environment is embedded in the myriad behaviors
and adaptations expressed as life history diversity in populations
(Levin 1992). Changing habitat configurations in dynamic
and easily fragmented stream environments (Fagan 2002) may
have contributed to the variety of life history strategies found
within and among native salmon and trout species (Rieman
and Dunham 2000).
Resilient native salmon and trout must be able to persist
in stream networks in which topology changes over time
10
and across space. Much of the life history diversity among
and within populations of salmon and trout focuses on
movement. Dynamic network topology allows consideration
of the adaptive significance of individuals that move compared
to individuals that do not move. The complement of
behavioral variation expressed as movement by individuals in
a population may increase chances of population persistence
in uncertain and dynamic stream network environments.
For example, some individual trout will move great distances
in their lifetime, while others will stay close to their natal
pool (Heggenes et al. 1991). Both of these life history types
will be most successful with different sets of environmental
conditions as reflected by network juxtaposition and access to
habitats. In years when adequate seasonal habitats for the trout
are close to one another, both movement types will survive. In
years when seasonal habitats are far apart or when disturbance
makes network habitats inhospitable, the trout that moves the
most may be more likely to survive than the trout that does
not move.
The adaptive importance of movement behavior for
survival (Levin 1992) in a dynamic stream network may
warrant explorations of network topology that are species or
life history stage specific. For example, the two freshwater life
history stages (juvenile and spawning adult) of coho salmon
(O. kisutch) have different mobility and life history needs that
define the extent of the stream network that individuals at
each life history stage will explore and occupy. Individual adult
coho salmon are seeking spawning habitat and mates, and may
travel among tributaries or even subbasins before spawning.
Juvenile coho salmon must survive in freshwater for one year,
therefore, individuals must find seasonal habitats within their
natal stream system. Therefore, adults may respond to the
network configuration of spawning habitats, while juvenile
distribution may be associated with the juxtaposition of
the several habitat types necessary for freshwater residency.
These two sets of needs may overlap on a stream network, or
they may not. It is possible that adults that are only seeking
spawning habitat and mates would spawn in places that are
not close to juvenile rearing habitat. Although the propensity
of adult salmon to spawn in natal streams may compensate for
this, presumably because the adult was able to rear in that site
as a juvenile, this may not always be the case. Assessments of
population viability that are based on instream habitat may
be improved by considering the proximity between habitats
necessary for each life history stage.
Networks and Spatial Scale
Changing topology has different effects at different levels
of organization, from the scale of individual sites (e.g., pools)
to stream reaches and entire watersheds (i.e., Burnett 2001,
Church 2002). Domains of scale exist in ecology, and these
describe the observation that patterns found at one spatial or
Flitcroft
Chapter 2
temporal scale may be undetectable or look entirely different
at another scale (Wiens 1989). For example, patterns detected
at small spatial extents may reflect variation in the choices of
individuals, while patterns over large extents may be associated
with overall population level variation. The effect that spatial
or temporal scale has on our comprehension of pattern
associations in a stream network has rarely been explored by
researchers. This may reflect the static or linear view of stream
systems that results in analysis of disconnected sites in stream
networks rather than continuous data collection throughout
the network. Also, collecting comprehensive census
information for a network is time consuming and costly, and
few tools are available to analyze data that incorporate a stream
network framework (see Analysis and Dynamic Network
Topology section).
Ultimately, in many aquatic investigations, the nuances
of scale and stream network relationships are lacking in the
interpretation of patterns. This could result in misinterpretation
of the meaning of spatial patterns of habitat, occupancy, or use
within a stream network. For example, there is much debate
about the effectiveness of instream habitat enhancement
structures for juvenile salmonids. Studies have shown that
such structures will result in greater numbers of juveniles at
sites than were present before the structures were in place.
Biologists assume that this means that juvenile survival has
increased due to the structures. However, no research has
attempted to identify where the juveniles came from that
are now inhabiting the structure. They could simply have
moved to the site from up or downstream. Or they could be
emerged juveniles from the site itself. The pattern of juvenile
occupancy has certainly changed within the stream system,
but by only sampling at the site where the restoration was
done, it is not possible to determine the overall effect of the
habitat enhancement structures on the population of juveniles
throughout the stream network.
A network perspective that includes dynamic network
topology facilitates relevant ecological interpretations of
patterns detected at multiple spatial scales by providing a
framework within which to consider movement paths, habitat
juxtaposition, and specific life history stage requirements.
Flitcroft (2007) incorporated different network variables at
several spatial scales to explore patterns in the distribution
of juvenile coho salmon (O. kisutch). Juvenile coho salmon
density, explored at the site scale, pointed to the importance
of the close proximity of seasonal habitats among subbasins
whose juvenile density changed between summers. Patch
sizes were identified by variograms that used stream network,
rather than Euclidean, distances between points. Patterns of
juvenile clumping identified as patches appeared to reflect
subbasin scale variation. Dynamic network topology provided
a framework within which to consider local site conditions
through measures of habitat juxtaposition and large scale
11
subbasin conditions that may be associated with environmental
changes over time.
Spatial scale and the consideration or interpretation of
occupancy patterns and movement in the stream network may
be further complicated by considering interactions among the
life history stages of a species. Consider, for example, juvenile
and adult anadromous coho salmon. As previously mentioned,
these two life history stages of coho salmon share the same
distribution in the stream network but have different habitat
needs and mobility. Flitcroft (2007) found that network
occupancy by juvenile coho salmon appeared to be related to
adult spawning run sizes. In years of large adult run sizes, more
of the stream network was occupied by juveniles. Therefore,
the pattern of juvenile coho salmon network occupancy is
mediated by the distribution pattern of spawning adults that
were not necessarily seeking habitats necessary for juvenile
survival. Analyses of local patterns of juvenile distribution that
fail to include a consideration of larger scale patterns of adult
run size may misrepresent the importance of local habitat
conditions.
Isolation and Movement in Stream Networks
The ability to move among habitats is critical for survival of
native aquatic organisms (Hanski and Gilpin 1991, Gresswell
et al. 2006). The concept of isolation applies to habitats or
individuals. Habitats may be isolated from one another due
to the configuration of the stream network or environmental
conditions (i.e., disturbance, discharge). Individuals may be
isolated from one another or from the habitats that they need
to survive. Isolation for individuals in either of these instances
may be due to the physical configuration of the stream
network, or to the level of mobility of the individual. For
example, a species of salamander with limited mobility will
need habitat patches that are close together to avoid isolation
from necessary habitats. This compares to a highly mobile
trout that may survive even when necessary habitats are
separated by many kilometers of stream channel. Therefore,
habitat topology necessary for the salamander to survive will
be different than the topology for the trout, and not simply
because they have different habitat needs. The implications
of dynamic habitat changes due to disturbance or seasonal
discharge will apply differently to species depending on
mobility. This is ecologically important when assessments
of stream habitat are part of population viability analyses.
Analyses that do not consider species-specific habitat and the
mobility of individuals may not accurately gauge the amount
of habitat available for long-term persistence.
Isolation within and among stream networks may be a
significant feature describing the distribution of a population
of organisms. For example, in the arid mountain west of
the United States, stream systems and their populations of
Flitcroft
Chapter 2
species may become isolated seasonally, and some systems
are permanently separated from one another, due to longterm climate change. In these arid environments, population
viability assessments that span isolated stream systems
should take into account the lack of connectivity between
subpopulations. Consider redband trout (Oncorhynchus mykiss)
that are distributed across the interior Great Basin of Oregon.
Populations of trout are permanently isolated from one
another in stream networks that were once connected as part
of a large lake. Further, there is seasonal isolation of habitats in
the Great Basin when low water summer months result in dry
stream beds and high temperature stream reaches that may be
impassable to redband trout. The lack of connectivity among
streams means that there is no opportunity for a stream whose
fish have been extirpated (due perhaps to a disturbance event
or high stream temperatures) to be recolonized and recover
to pre-extirpation levels. Current population assessments
of redband trout (Jones et al. 2007) discuss the isolation of
subpopulations in disconnected basins, but do not consider
the lack of connectivity among or within stream systems in the
anticipated long term survival of redband trout. By failing to
consider system connectivity and the effects of disturbances in
fragile desert streams, population assessments based simply on
population estimates without considering the vulnerability to
extinction of isolated subpopulations may exaggerate a species’
anticipated long-term persistence.
Analysis and Dynamic Network Topology
How variables describing stream conditions are analyzed
is an ongoing challenge to stream ecologists. The framework
of dynamic network topology does not include new network
based statistics for analysis. However, considering current
statistical tools that include stream networks, and evaluating
traditional statistical tools with regards to network relationships
is an important element of any approach that is meant to
contribute to a network-based ecological investigation.
Independence of observations is a fundamental assumption
of many parametric statistical methods. Lack of independence,
which includes correlation among observations, may mean
that one observation contains some of the same information
that is contained in other observations. Ignoring correlation
can produce incorrect estimates and standard errors. Therefore,
standard statistical methods for analyzing stream data may
be compromised by the inherent connectivity of a stream
network.
Often, researchers simply include models for correlated
error terms in regression models, in an attempt to capture the
lack of independence among sample sites in stream systems
(i.e., Megalhaes et al. 2002, Joy and Death 2004, Grenouillet
et al. 2004). Such work has indicated that stream network
criteria can be critically important in understanding the
12
patterns of aquatic biota. However, developing and using
variables that reflect both network proximity and habitat
types without reducing analysis to a comparison of network
distances is challenging.
There have been some attempts at developing statistical
sampling protocols that allow researchers to capture the
variation in an entire population of streams (Larsen et al.
2001, Jacobs and Cooney 1995). Such techniques focus on
sampling designs that allow for a spatially random selection
of sample sites across a population of streams. This type of
sampling design effectively describes the mean and variance of
individual stream variables over a large region, but how clearly
this reflects the subtle pulse and flow of contiguous stream use
by populations of organisms continues to be an open question.
Also, researchers must decide for themselves whether seeking
to develop sampling strategies and analyses that negate the
correlation inherent in most network based systems is the most
ecologically intuitive approach. Statistical methods that take
advantage of correlation, rather than viewing correlation as a
nuisance, may more realistically reflect ecological systems and
better inform our understanding of the interconnectedness of
fluvial systems and the organisms that depend on them.
In terrestrial ecology, spatial statistical methods have
been developed that take advantage of autocorrelation by
considering the proximity between observations that is
inherent in the landscape (Legendre and Fortin 1989). These
statistical techniques allow researchers to explore and compare
patterns across the landscape. Unfortunately for aquatic
ecologists, such spatial statistics are generally inappropriate for
work with stream systems, due to directionality of water and
nutrient flow. Directionality is not a consideration of spatial
statistics that seek shared variation among points in every
direction, not just upstream or downstream, in order to detect
patterns in a dataset.
Pioneering work has been done recently that attempts to
modify spatial statistical tools (Ganio et al. 2005, Cressie
et al. 2006, VerHoef et al. 2006) or analysis (Gresswell et
al. 2006) to the strictures of stream environments. Because
there is a direction of flow in stream systems, statistics may
develop that are similar to the flow of time in time series data:
a point in the past can influence a point in the future but
not vice versa. Spatial statistical methods in streams would
allow influence to flow directionally between two points.
Aquatic species must be considered within the context of the
stream network, and within the framework of their ability
to move to areas upstream or downstream when patterns are
described and interpreted using statistics. The current work
with spatial statistics in streams that encompasses upstream or
downstream directionality is promising and may well lead to
analytical tools that more closely reflect the movement options
of individual fish.
Flitcroft
Chapter 2
An Application of Dynamic Network Topology
The application of dynamic network topology may best
be explained using an example. Consider coho salmon. As
previously mentioned, these fish use freshwater streams in
their adult and juvenile life history stages, and fish in these two
life stages have different levels of mobility and habitat needs.
In the Pacific Northwest, much time, energy, and expense has
been spent on coho salmon habitat enhancement projects.
However, these projects do not consider the juxtaposition of
habitats for each life history stage. By including considerations
of the proximity and quality of existing habitats, enhancement
work may be more effective at accomplishing its goal of
providing more habitat for coho salmon.
In many habitat assessments for coho salmon, the number
of stream kilometers of certain types of stream habitat is
summarized at a subbasin or basin scale or habitats for
spawning adults, summer juvenile rearing pools, and winter
off channel refuge habitats for juveniles are often calculated
(i.e., Johnson 1999). For example, in subbasin “A” there
might be 2 km of spawning beds associated with 5 spawning
reaches, 4 km of summer rearing habitat corresponding with
8 pools, and 3 km of winter habitat corresponding with 4
off-channel habitats. This combination of habitats in a basin
may be deemed adequate to sustain the population of coho
salmon that live there. This may be the case if the habitats are
intermixed and do not require great navigation in order for
the fish at each life stage to find the necessary habitats. From a
dynamic network topology perspective, summary information
that might describe juxtaposition would include the proximity
along the network between spawning habitat and summer
pools, and then between summer pools and winter rearing
Figure 2.5. Two extreme examples
of habitat arrangements may
be conceptually useful when
considering how an individual
juvenile coho salmon must
move within the network to find
necessary seasonal. Figures (a) and
(b) share the same stream network,
and the same numbers of habitat
sites classified for coho salmon
as spawning (S), summer refuge
(sum), or winter rearing (win).
However, the arrangement of the
habitats is quite different with
figure (a) depicted interspersion of
habitats while each seasonal habitat
is restricted to specific tributaries
in (b).
13
habitats. This would be the progression between habitats that
juvenile coho salmon would need.
Consider two different orientations of habitat in subbasin
“A” that might depict extreme examples of habitat organization.
In the first configuration, habitats are intermixed, and network
distances between seasonal habitats are short (Figure 2.5a).
In the second configuration, each set of habitats is isolated
in individual tributaries, complicating navigation between
seasonal habitats through the network, and also causing
distances between habitats to be large (Figure 2.5b). Although
both scenarios contain the same quantities of habitat, the
implications of the configuration of habitats for juvenile
survival are different. In the intermixed subbasin, juveniles
may have little difficulty finding necessary habitats, but
juveniles in subbasins with habitats isolated from one another
may have difficulty finding particular habitats. Isolation
of habitats may possibly result in lower juvenile survival or
growth. If a watershed manager was faced with these two sets
of subbasin configurations, targeting restoration projects that
provide seasonal habitats in a pattern that reduces navigation
difficulty or network distances might be the most successful
course of action.
Another element of dynamic network topology is the
landscape context, which affects how habitats will change over
time. Consider the two subbasins described previously. If these
two subbasins have the same underlying geology and land use,
it might be efficient for a watershed manager to consider how
the landscape may affect the stream in the future. For example,
if a portion of the watershed was in an agricultural classification,
it is possible that erosion in these areas would be detrimental
to juvenile survival. Hence, restoration that encouraged
juveniles to live in these areas would be counterproductive
Flitcroft
Chapter 2
in the future. Rather, focusing work in areas where future
disturbance events may deliver material necessary for habitat
creation may be useful. For example, focusing on restoration
in areas that have large trees in the riparian zone that may be
expected to be delivered to the stream in future decades might
be a more productive location to encourage fish to populate.
Each watershed will have a unique configuration of habitats
and potential for disturbances. However, dynamic network
topology may provide a way of thinking about the current and
future conditions of the stream network from the perspective
of the aquatic species whose survival is being analyzed.
Dynamic network topology is not meant to describe a
process of analysis as much as a framework from which to
approach population level analysis of habitats in stream basins.
Considering the topology of habitats and the effect that
change will have in a watershed may be useful in developing
management that seeks to restore populations of aquatic
organisms.
Conclusions
With the accepted awareness by ecologists of the
importance and fundamental place of disturbance in creating
and maintaining ecosystems, new ways of thinking about
organization are necessary. In aquatic ecology, a static or
linear perspective is being replaced by a focus on dynamic
processes. The need for a network perspective is becoming
increasingly clear, as research shifts from a site-level focus to
the consideration of watersheds and the movement patterns of
populations of aquatic species. Dynamic network topology is
a universal notion that can inform assessments of population
viability, habitat quality and quantity, and the interrelationship
between aquatic species and the environment. Dynamic
network topology may help aquatic biologists understand
patterns of occupancy, dispersal, and life history needs within
the context of a stream network.
This paper is intended to raise the awareness of aquatic
ecologists about the need for consideration of species within
the framework of an interconnected stream environment. I
propose the framework of dynamic network topology as a
conceptual base from which to consider stream processes and
the needs of aquatic species. Further research that incorporates
movement paths, fragmentation, and isolation in a stream
network would help deepen our understanding of the ecology
and needs of aquatic species. Including spatial and temporal
scale in the interpretation of patterns of occupancy and
movement must also be considered. It is of critical importance
for the conservation of aquatic obligate stream species that
biologists develop an adequate working understanding of the
mechanisms of dispersal and habitat use in the stream network.
The majority of examples and ideas that have informed
the creation of a dynamic network topology come from
Pacific salmon and trout. However, the concept that the
14
configuration of stream network habitats changes over time,
and that native species are adapted to cope with these changes
may be applied to any species with an aquatic obligate life
history stage. The complexity of habitat juxtaposition in
stream networks, coupled with limited directions of movement
due to network configuration and directionality of flow, are
implicit in any population level analysis of an aquatic species.
Further consideration and discussion of network connectivity,
movement options, and habitat juxtaposition will enrich and
define aquatic ecological theory and practice.
Literature Cited
Benda, L., D. Miller, T. Dunne, J. Agee and G. Reeves. 1998.
Dynamic landscape systems, Chapter 12. Pages 261-288 in R.J.
Naiman and R.E. Bilby, editors. River ecology and management:
lessons from the Pacific Coastal Ecoregion. Springer-Verlag, New
York.
Benda, L., L. Poff, D. Miller, T. Dunne, G. Reeves, G. Pess and M.
Pollock. 2004. The network dynamics hypothesis: how channel
networks structure riverine habitats. BioScience 54(5): 413-427.
Burnett, K.M. 2001. Relationships among juvenile anadromous
salmonids, their freshwater habitat, and landscape characteristics
over multiple years and spatial scales in the Elk River, Oregon.
Doctoral dissertation. Oregon State University, Corvallis, Oregon.
Chapman, D.W. 1962. Food and space as regulators of salmonid
populations in streams. The American Naturalist 100(913): 345357.
Church, M. 2002. Geomorphic thresholds in riverine landscapes.
Freshwater Biology 47: 541-557.
Cooney, C.X. and S.E. Jacobs. 1995. Oregon coastal salmon
spawning surveys, 1993. Oregon Department of Fish and Wildlife,
Fish Division, Portland, Oregon.
Cressie, N., J. Frey, B. Harch and M. Smith. 2006. Spatial prediction
on a river network. Journal of Agricultural, Biological and
Environmental Statistics 11(2): 127-150.
Fagan, W.F. 2002. Connectivity, fragmentation, and extinction risk
in dendritic metapopulations. Ecology 83(23): 3243-3249.
Fausch, K.D., C.E. Torgersen, C.V. Baxter and H.W. Li. 2002.
Landscapes to riverscapes: bridging the gap between research and
conservation of stream fishes. BioScience 52(6): 483-498.
Fisher, S.G. 1997. Creativity, idea generation, and the functional
morphology of streams. Journal of the North American
Benthological Society 16: 305-318.
Flitcroft, R.L. 2007. Chapter 3. This dissertation. Oregon State
University, Corvallis, Oregon.
Frissell, C.A., W.J. Liss, C.E. Warren and M.D. Hurley. 1986. A
hierarchical framework for stream habitat classification: viewing
streams in a watershed context. Environmental Management
10(2): 199-214.
Flitcroft
Chapter 2
Ganio, L.M., C.E. Torgersen and R.E. Gresswell. 2005. A
geostatistical approach for describing spatial pattern in stream
networks. Frontiers in Ecology and the Environment 3(3): 138144.
Gardner, B., P.J. Sullivan and A.J. Lembo, Jr. 2003. Predicting
stream temperatures: geostatistical model comparison using
alternative distance metrics. Canadian Journal of Fisheries and
Aquatic Sciences 60: 344-351.
Gerking, S.D. 1959. The restricted movement of fish populations.
Biological Review 34: 221-242.
Gleason, H.A. 1926. The individualistic concept of the plant
association. Torreya 53: 7-26
Gleason, H.A. 1927. Further views of the succession concept.
Ecology 8: 299-326.
Grenouillet, G., D. Pont and C. Herisse. 2004. Within-basin fish
assemblage structure: the relative influence of habitat versus
stream spatial position on local species richness. Canadian Journal
of Fisheries and Aquatic Sciences 61: 93-102.
Gresswell, R.E., C.E. Torgersen and D.S. Bateman. 2006. A
spatially explicit approach for evaluating relationships among
coastal cutthroat trout, habitat, and disturbance in headwater
streams. Pages 457-471 in R.M. Hughes, L. Wang and P.W.
Seelbach, editors. Influences of landscapes on stream habitats and
biological assemblages. American Fisheries Society, Symposium
48, Bethesda, Maryland.
Hanski, I. and M. Gilpin. 1991. Metapopulation dynamics: a brief
history and conceptual domain. Biological Journal of the Linnean
Society 42: 3-16.
Heggenes, J., T.G. Northcote and A. Peter. 1991. Spatial stability of
cutthroat trout (Oncorhynchus clarki) in a small, coastal stream.
Canadian Journal of Fisheries and Aquatic Sciences 48: 757-762.
Horn, H.S. 1981. Chapter 11: Succession. Pages 253-271 in May,
R.M. editor. Theoretical Ecology: Principles and Applications.
Blackwell Scientific Publications, Boston.
Jacobs, S.E. and C.X. Cooney. 1995. Improvement of methods used
to estimate the spawning escapement of Oregon coastal natural
coho salmon. Oregon Department of Fish and Wildlife, Portland,
Oregon.
Jones, K.K., J.M. Dambacher and R.L. Flitcroft. 2007. Effectiveness
and applicability of EMAP survey design in status review of Great
Basin redband trout. Extended Abstract to be published as part of
2007 Redband Workshop.
Joy, M.K. and R.G. Death. 2004. Predictive modeling and spatial
mapping of freshwater fish and decapod assemblages using GIS
and neural networks. Freshwater Biology 49: 1036-1052.
Kahler, T.H., P. Roni and T.P. Quinn. 2001. Summer movement
and growth of juvenile anadromous salmonids in small western
Washington streams. Canadian Journal of Fisheries and Aquatic
Sciences 58: 1947-1956.
Kocik, J.F. and C.P. Ferreri 1998. Juvenile production variation in
salmonids: population dynamics, habitat, and the role of spatial
relationships. Canadian Journal of Fisheries and Aquatic Sciences
55(Suppl. 1): 191-200.
15
Landres, P.B., P. Morgan and F.J. Swanson. 1999. Overview of the
use of natural variability concepts in managing ecological systems.
Ecological Applications 9(4): 1179-1188.
Larsen, D.P., T.M. Kincaid, S.E. Jacobs and N.S. Urquhart. 2001.
Designs for evaluating local and regional scale trends. BioScience
51(12): 1069-1078.
Legendre, P. and M.J. Fortin. 1989. Spatial pattern and ecological
analysis. Vegetatio 80: 107-138.
Levin, S.A. 1992. The problem of pattern and scale in ecology: The
Robert H. MacArthur award lecture. Ecology 73(6): 1943-1967.
Li, W.H. and twelve coauthors. 1995. Safe havens: refuges and
evolutionarily significant units. Pages 371-380 in J.L. Nielsen,
editor. Evolution and the aquatic ecosystems: defining unique
units in population conservation. American Fisheries Society,
Symposium 17, Bethesda, Maryland.
Lowe, W.H., G.E. Likens and M.E. Power. 2006. Linking scales in
stream ecology. BioScience 56(7): 591-597.
Megalhaes, M.F., D.C. Batalha and M.J Collares-Pereira. 2002.
Gradients in stream fish assemblages across a Mediterranean
landscape: contributions of environmental factors and spatial
structure. Freshwater Biology 47: 1015-1031.
Naiman, R.J., T.J. Beechie, L.E. Benda, D.R. Berg, P.A. Bisson,
L.H. MacDonald, M.D. O’Connor, P.L. Olson and E.A. Steel.
1992. Fundamental elements of ecologically healthy watersheds
in the Pacific Northwest coastal ecoregion. Pages 127-187 in R.J.
Naiman, editor. Watershed Management: Balancing Sustainability
and Environmental Change. Springer-Verlag, New York.
Nakamura, F., F.J. Swanson and S.M. Wondzell. 2000. Disturbance
regimes of stream and riparian systems – Adisturbance-cascade
perspective. Hydrological Processes 14: 2849-2860.
Nielsen, J. 1992. Microhabitat-specific foraging behavior, diet, and
growth of juvenile coho salmon. Transactions of the American
Fisheries Society 121: 617-634.
Ottaway, E.M. and D.R. Forrest. The influence of water velocity
on the downstream movement of alevins and fry of brown trout,
Salmo trutta L. Journal of Fish Biology 23(2): 221-227.
Pacific Fisheries Resource Conservation Council. 1998. Biological
stock structure: the staggering diversity in salmon population.
Salmon Stocks Background Paper No. 1999/1b.
Poff, N.L., J.D. Allan, M.B. Bain, J.R. Karr, K.L. Prestegaard, B.D.
Richter, R.E. Sparks and J.C. Stromberg. 1997. The natural
flow regime a paradigm for river conservation and restoration.
BioScience 47(11): 769-784.
Quinn, T.P. 2005. The behavior and ecology of Pacific salmon and
trout. University of Washington Press, Seattle, Washington.
Reeves, G.H., L.E. Benda, K.M. Burnett, P.A. Bisson and J.R. Sedell.
1995. A disturbance-based ecosystem approach to maintaining
and restoring freshwater habitats of evolutionarily significant units
of anadromous salmonids in the Pacific Northwest. American
Fisheries Society Symposium 17: 334-349.
Flitcroft
Chapter 2
16
Reeves, G.H., K.M. Burnett and S.V. Gregory. 2002. Fish and
aquatic ecosystems of the Oregon Coast Range. in Hobbs, S.D.,
J.P. Hayes, R.L. Johnson, G.H. Reeves, T.A Spies, J.C. Tappeiner
II and G.E. Wells, editors. Forest and Stream Management in the
Oregon Coast Range. Oregon State University Press, Corvallis,
Oregon.
Torgersen, C.E., R.E. Gresswell and D.S. Bateman. 2004. Pattern
detection in stream networks: Quantifying spatial variability in fish
distribution. In T. Nishida, P.J. Kailola and C.E. Hollingworth,
eds. GIS/Spatial Analyses in Fishery and Aquatic Sciences, pp.
405-420, Fishery and Aquatic GIS Research Group, Saitama,
Japan.
Richards, C. and P.J. Cernera. 1989. Dispersal and abundance
of hatchery-reared and naturally spawned juvenile Chinook
salmon in an Idaho stream. North American Journal of Fisheries
Management 9: 345-351.
Vannote, R.L., G.W. Minshall, K.W. Cummins, J.R. Sedell and C.E.
Cushing. 1980. The river continuum concept. Canadian Journal
of Fisheries and Aquatic Sciences 37: 130-137.
Rieman, B.E. and J.B. Dunham. 2000. Metapopulations and
salmonids: a synthesis of life history patterns and empirical
observations. Ecology of Freshwater Fish 9: 51-64.
Scott, J.M., B. Csuti, J.D. Jacobi and J.E. Estes. 1987. Species
richness: A geographic approach to protecting future biological
diversity. BioScience 37(11): 782-788.
Southwood, T.R.E. 1977. Habitat, the templet for ecological
strategies? The Journal of Animal Ecology 46(2): 336-365.
Stiling, P.D. 1992. Introductory ecology. Prentice Hall, Englewood
Cliffs, New Jersey.
Thurow, R.F., D.C. Lee and B.E. Rieman. 1997. Distribution and
status of seven native salmonids in the interior Columbia River
basin and portions of the Klamath River and Great Basins. North
American Journal of Fisheries Management 17: 1094-1110.
VerHoef, J.M., E. Peterson and D. Theobald. (2006). Spatial
statistical models that use flow and stream distance. Environmental
and Ecological Statistics 13: 449-464.
Wallington, T.J. 2005. Implications of current ecological thinking
for biodiversity conservation: A review of the salient issues.
Ecology and Society 10(1): 15.
Warren, C.E. and W.J. Liss. 1980. Adaptation to aquatic
environments. Pages 15-40 in R. T. Lackey and L. Nielsen,
editors. Fisheries Management. Blackwell Scientific Publications,
Oxford, UK.
Wiens, J.A. 1989. Spatial scaling in ecology. Functional Ecology 3:
385-397.
Rebecca L. Flitcroft. 2007. Regions to streams: Spatial and temporal Variation in stream occupancy patterns of coho salmon (Oncorhynchus
kisutch) on the Oregon Coast. Ph.D. dissertation. Department of Fisheries and Wildlife, Oregon State University, Corvallis, OR.
CHAPTER 3
Exploring Network Patterns in the Habitat, Abundance and Distribution of
Juvenile Coho Salmon (Oncorhynchus kisutch) in Oregon’s Mid-Coast
Rebecca L. Flitcroft
Department of Fisheries and Wildlife
Oregon State University
Corvallis, Oregon
Abstract
Analyzing streams as networks may offer opportunities to understand the movement and distribution of aquatic
species. Network analysis is challenging for aquatic ecologists due to limited research techniques and the need for
census style field data collections. Further, a stream network may provide opportunities to explore multiple spatial scales
within a stream system and identify ecological relationships that are scale specific. Studies that examine the implications
of network patterns at different spatial scales may contribute new ideas about aquatic species persistence and habitat
needs. In this study, the distribution of juvenile coho salmon in the stream network was investigated in eleven subbasins
of Oregon’s Mid-Coast at three spatial scales (site, patch and subbasin). At the site scale, a comparison of traditional
instream, site-specific habitat descriptors with network orientation and habitat juxtaposition variables was completed
in order to explore the usefulness of both types of metrics in understanding juvenile stream use. Variograms using
network distances were generated to describe patches of juvenile coho salmon. Patch size was compared to basin shape
to explore whether underlying geomorphology effected how juvenile coho clumped around habitat. The subbasin scale
allowed for the exploration of large-scale environmental conditions that may be important for of juvenile coho salmon
distribution (adult spawning run size and precipitation). A variety of statistical tools including principle components
analysis, discriminant analysis and a local-mean non-parametric multiplicative regression analysis were used to explore
connections between juvenile density and explanatory variables at each spatial scale. At the site scale, modeling and
principle components analyses were interpreted to mean that a diversity of habitats in close network proximity is
necessary for juvenile survival. Patches of juvenile coho were linked to the size of the watershed, and at a subbasin scale
it appeared that juvenile network occupancy was tied to the size of the spawning adult run. Each spatial scale identified
different relationships between juvenile coho and their environment and could be integrated to more fully describe
how juvenile coho use the stream network. Using an analytical perspective that includes stream network relationships
contributed to a meaningful interpretation of the pattern of juvenile occupancy, and conclusions about habitat needs at
this life history stage.
Introduction
Stream systems in the Pacific Northwest are characterized
by habitat and environmental conditions that may change
dramatically over time. Dynamic stream processes, including
disturbance (Nakamura et al. 2000), fire (Miller et al. 2003)
and flooding (Benda and Dunne 1997, Swanson et al. 1998),
work to connect or disconnect portions of the stream system.
In the past century, additional disturbance from anthropogenic
activities such as timber harvest, road building (Trombulak and
Frissell 2000), farming and urbanization (Lichatowich 1999)
have changed the extent and interval between disturbance
events (Reeves et al. 1995).
A dynamic stream environment may lead to a variety of
habitat types within the stream channel. Stream channels
are organized as a network of interconnected tributaries.
Ultimately, the stream network is the template upon which
landscape processes act and the resulting stream mosaic will
include a variety of habitat conditions (Naiman et al. 1992).
Therefore, the consideration of stream habitat diversity and
connectivity for aquatic species may benefit from including
the topology or physical orientation of habitats within a
stream network.
Flitcroft
Chapter 3
How aquatic obligate species respond to the configuration of
habitats in a stream network may provide insight into behaviors
or adaptations that have led to resilience in native organisms
(Flitcroft 2007). Fagan (2002) showed that fragmentation in a
stream system reduces connectivity among the different parts
of the network making navigation among necessary habitats
for aquatic species difficult. Fagan speculated that lack of
connectivity and isolation of habitats and individuals could
increase chances of biological extinction. Spatial location
of stream features has been explored in assessments of fish
assemblages (Megalhaes et al. 2002) and biotic diversity
(Day et al. 1992, Wright and Li 2002) in attempts to link
geomorphic patterns with biotic distributions. Montgomery
et al. (1999) connected stream channel type with spawning
distributions of salmonids and concluded that the size of
spawning fish determined redd depth, thereby dictating the
lowest range within the stream network that redds could be
dug and not disturbed by annual river flow and the ability of
rivers to move substrate.
An important research question facing modern ecologists
is: “What are the implications of network patterns of habitat,
occupancy and abundance for the long- term persistence of
aquatic obligate species?” Network patterns lend themselves
to analysis at multiple spatial scales. The network itself is one
large scale, and the sites that compose the samples are another.
Currently, ecologists recognize the need for analyses that
explore broad spatial and temporal scales that may contribute
to a better understanding of the needs of populations of species
(Fausch et al. 2002, Frissell 1986, Schlosser 1991, Talley 2007).
Burnett (2001) explored the use of different spatial extents in
describing the distribution of juvenile salmonid assemblages
in the Elk River Basin, Oregon. She found that spatial extent
over time was important in describing the observed variability
in assemblage distribution. However, much of our current
understanding about the distribution and needs of stream
fishes is derived from studies that consider relatively small
spatial extents, generally habitat units or reaches, without the
context of time, multiple spatial scales or network connectivity
(Matthews and Marsh-Matthews 2003). Small scale studies
are not able to detect patterns at larger scales (Wiens 1989)
and may be unable to detect patterns of connectivity in an
entire network. Studies that include network connectivity
at multiple spatial scales may allow ecologists to learn about
ecological processes that contribute to the persistence and
resilience of native species.
Although many ecologists recognize the importance of
spatial scale and stream connectivity, incorporating such a
perspective continues to be challenging. In Chapter 2 (Flitcroft
2007), I explored why a network perspective has been elusive in
the discipline of aquatic ecology. Two reasons I identified were
the lack of research of domains of scale (patterns at different
scales describe different processes, [Wiens 1989]) in aquatic
18
ecology and the paucity of statistical methods developed for
stream network applications. Analysis incorporating a network
framework is difficult due to the lack of research on what
patterns of stream fishes mean at multiple scales and network
configurations and is compounded by a lack of independence
in stream networks that violates assumptions of parametric
statistics making any analysis challenging.
Aquatic obligate species such as Pacific salmon
(Oncorhynchus spp.) face unique challenges living in and using
multiple locations within a stream network. Individuals must
be able to migrate into and out of a stream system and find
appropriate habitats at each life history stage. The proximity
and connectivity of habitats may be a significant consideration
when trying to understand the distribution and abundance
of salmonids (Benda et al. 2004). Coho salmon (O. kisutch)
provide an opportunity to explore patterns of network use
due to their broad distribution and the depth of research that
has focused on their life history expression and needs. As an
anadromous species, sexually mature adults home to their
natal streams to spawn and then die (Groot and Margolis
1991). Some adult coho salmon stray rather than returning
to their natal stream (Labelle 1992). Before migrating to the
ocean, juveniles spend a year rearing in freshwater and may
use different seasonal habitats that are located throughout a
watershed (Nickelson et al. 1992a, Nickelson et al. 1992b,
Sleeper 1993).
A network framework incorporated into an analysis of
multiple spatial scales over time may provide insights into
juvenile coho salmon occupancy, distribution and abundance.
In this retrospective study, three objectives were identified that
explore how a stream network framework may be incorporated
into an analysis of juvenile coho over a five-year period across
multiple stream systems and spatial scales. My first objective
was to compare the explanatory power of instream and network
habitat variables for each year of juvenile coho salmon density.
The second objective was to characterize the size of habitat
patches occupied by juvenile coho salmon and examine
relationships with basin shape and area. The third objective was
to describe juvenile coho salmon network occupancy in eleven
subbasins and consider large-scale environmental conditions
that might be important for explaining occupancy patterns.
Due to the challenge of collecting multiple pre-existing
datasets for this analysis, different objectives are explored with
different combinations of subbasins.
Materials and Methods
Study Area
Subbasins in the mid-coast of Oregon (Figure 3.1) were
selected for analysis because several organizations (i.e., Oregon
Department of Fish and Wildlife and the Mid-Coast Watershed
Flitcroft
Chapter 3
19
Figure 3.1. Highlighted subbasins in the Alsea
and Siletz River Basins that were included in an
analysis of multiple spatial scales that incorporated
a stream network framework.
Council) have studied adult and juvenile life history stages of
coho salmon in this area. The mid-coast of Oregon is an area
of similar geology, vegetation, land use and precipitation. The
headwaters of the larger Siletz and Alsea River basins begin
in the Oregon Coast Range and are covered predominantly
by conifer forests. Timber harvest is an important economic
activity in the higher elevation areas and agriculture is the
dominant land use activity in the lowlands. The geology is
a mixture of volcanic and sedimentary rocks, and volcanic
terrain generally found to the north. The climate of this region
is mild with precipitation occurring predominantly in the
winter months (Redmond and Taylor 1997).
Some populations of coho salmon are listed as endangered by
the state of Oregon (ODFW 2005). This species has significant
commercial and recreational value for communities on the
Oregon coast. However, the stream habitat of coho salmon has
been heavily affected by anthropogenic disturbances that have
increased sediment loads, decreased wood content of streams,
and introduced barriers to migration such as culverts and
water diversions (Lichatowich 1999). Juvenile coho salmon
over-winter primarily in coastal streams of relatively small
stream size (1st–3rd order on 1:100,000 scale maps, Strahler
1952) (Rosenfeld et al. 2000). Research and monitoring data
are available that document juvenile coho salmon distribution,
abundance, habitat, behavior and life history diversity.
Multiple spatial extents were incorporated into this analysis
(Figure 3.2). The site scale corresponds with the size of pools
where juvenile coho salmon were snorkeled and was used
to explore instream and network habitat variables in close
proximity to the pools. The patch scale varied between 100 and
5000 meters and described the size of clumps of juvenile coho
salmon. Patch size was compared to subbasin scale conditions.
The subbasin scale ranged in size from 32 to 300 km2 and
allowed for the characterization of subbasins with regard to
juvenile occupancy changes over the five years of this study.
Datasets
Juvenile coho salmon counts for the years 1998, 1999, 2001
and 2002 came from snorkel surveys conducted by the MidCoast Watershed Council. Survey data from 2000 were not
used because accurate field maps necessary for georeferencing
were unavailable. During a summer field season (June through
September), every fifth pool was snorkeled, and all fish were
counted from the stream mouth to the headwaters, including
all tributaries. Surveys ended for a particular stream when
five consecutive pools surveyed contained no juvenile coho
salmon. Although juvenile coho salmon may move during any
season of the year (Kahler et al. 2001), summer surveys are
important because this is the time of year when juveniles are
Flitcroft
Chapter 3
20
Figure 3.2. The three spatial scales included in a
multiple scale analysis of subbasins in the Alsea and
Siletz River basins are the site, patch and subbasin.
assumed to move the least (Nickelson et al. 1992b). Therefore,
summer snorkel surveys permitted collection of comparable
snorkel information over several months across years.
Adult coho salmon spawning run sizes for the mid-coast
of Oregon were taken from reports prepared by the Oregon
Department of Fish and Wildlife’s spawning survey program
(Jacobs et al. 2002). The spawning run size estimates were
calculated using randomly selected sample locations that were
revisited regularly by field crews during the spawning season
to count both spawning fish and redds (Jacobs and Nickelson
1989). The randomly collected data was then used to estimate
the size of the adult coho salmon spawning run.
Fine scale instream site specific habitat survey information
came from the Oregon Department of Fish and Wildlife
Aquatic Inventories Program (AIP). Field habitat survey crews
walked from the stream mouth to the headwaters during
the summer, gathering measurements of the physical stream
environment and identifying discrete habitat units such as
pools, riffles and glides (Moore et al. 1997). Field surveys
from 1997 through 2002 were pieced together to encompass
the subbasins of interest for this study. A total of seventeen
instream habitat metrics were considered in this analysis and
were transformed, as necessary, for normality (Table 3.1).
Information about geology was acquired from a dataset
created by the United States Geological Survey (USGS) that
contained digitized geology information at a 1:500,000 scale.
Mean annual precipitation for Oregon was compiled by the
PRISM (Parameter-elevation Regressions on Independent
Slopes Model) group at Oregon State University, part of the
Oregon Climate Service. Rainfall records between 1971 and
2000 were used in an assessment process that incorporated
spatial effects of rain shadows and coastal effects (Daly et al.
1999).
Generating Stream Network Variables
Stream network variables need to represent the complexity,
connection, and hierarchical organization of a stream. These
variables and the stream network itself were derived by the
Coastal Landscape Analysis and Modeling Study (CLAMS).
Stream lines were generated from 10-m digital elevation
models (Miller 2003, Clarke et al. in review). A geographic
information system (GIS) (ArcInfo version 9.1) was used to
overlay the modeled stream data from the CLAMS dataset,
site-specific survey information from the AIP project, and
the juvenile snorkel survey points. Stream network variables
that represented habitat interconnectedness and network
spatial organization included stream order, mean annual
precipitation (mm), mean gradient downstream (from the
stream segment under consideration), coho salmon intrinsic
potential (Burnett et al. 2007), gradient of intrinsic potential
for coho salmon, elevation, stream width and valley width
index. Because variables were on many different scales, some
variables were modified for normality using logarithmic or
square root transformations as necessary to ease modeling and
analysis (Table 3.1).
The network juxtaposition of critical seasonal habitats
(spawning, summer refuge and winter rearing) required the
identification of adequate seasonal habitats. Seasonal habitats
were queried from the ODFW AIP habitat surveys. The
criteria used to define “adequate” habitat were based on the
ODFW Benchmarks (Appendix 3.1) for site-specific habitat
Flitcroft
Table 3.1. Variables in
network and habitat datasets
with source information and
transformations.
Chapter 3
21
Variable
Transformations
Source
Dataset
Measurement Type
Network Dataset
Stream Order
Mean Annual Precipitation
Intrinsic Potential Coho
Intrinsic Potential Coho – Gradient
Stream Width
Valley Width Index
Elevation
Mean Gradient Downstream
Dist. to Spawning Habitat
Dist. to Winter Rearing Habitat
Dist. to Summer Habitat
None
None
None
None
Logarithmic
Logarithmic
Square Root
Logarithmic
None
None
None
CLAMS
CLAMS
CLAMS
CLAMS
CLAMS
CLAMS
CLAMS
CLAMS
CLAMS
CLAMS
CLAMS
Modeled
Modeled
Modeled
Modeled
Modeled
Modeled
Modeled
Modeled
Modeled
Modeled
Modeled
Habitat Dataset
Shade
Valley Form
Slope
Depth
Percent Silt and Organic Substrate
Percent Sand in Substrate
Percent Gravel in Substrate
Percent Cobble in Substrate
Percent Boulder in Substrate
Percent Bedrock in Substrate
Boulder Count
Active Channel Erosion
Percent Undercut Bank
Number of Pieces of Wood
Wood Volume
Key Pieces of Wood
Percent Canopy Closure
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
ODFW AQI
ODFW AQI
ODFW AQI
ODFW AQI
ODFW AQI
ODFW AQI
ODFW AQI
ODFW AQI
ODFW AQI
ODFW AQI
ODFW AQI
ODFW AQI
ODFW AQI
ODFW AQI
ODFW AQI
ODFW AQI
ODFW AQI
Field Estimated
Field Estimated
Field Estimated
Field Estimated
Field Estimated
Field Estimated
Field Estimated
Field Estimated
Field Estimated
Field Estimated
Field Estimated
Field Estimated
Field Estimated
Field Estimated
Field Estimated
Field Estimated
Field Estimated
Other Variables
Subbasin
Basin
None
None
Descriptive
Descriptive
Locator
Locator
features (Foster et al. 2001). Adequate spawning habitat
was defined as riffle units with ≥ 50 percent gravel and ≤ 8
percent silt. Summer habitat was defined as pools ≥ 0.5 m in
depth. Winter rearing habitat included all off-channel habitats
defined as backwater pools, alcoves or isolated pools. While
the importance of beaver pools as rearing habitat is well known
(Collen and Gibson 2001, Pollock et al. 2004), these habitats
were not included because they occurred in only one subbasin.
I did not want to misrepresent the importance of beaver pools
throughout the region. Also, in the subbasin that had beavers
present, there were alcoves and other off channel habitat
that represented available winter habitat. Each habitat unit
that met the criteria for adequate habitat was selected from
the main coverage and placed into a new coverage for that
type of habitat. The ArcView 3.2 (ESRI)1 extension, Shortest
Network Paths Version 1.1 (Remington 1999) was used to
calculate network distances between the juvenile snorkel sites
and seasonal habitats (Figure 3.3).
Network occupancy metrics were meant to represent the
amount of stream network in which juvenile coho salmon were
present at a subbasin scale. After the juvenile snorkel survey
data was plotted in a GIS, the number of stream kilometers
that encompassed the range of juvenile coho for each year, and
in each subbasin, could be calculated.
The use of trade or firm names in this publication is for reader
information and does not imply endorsement by the U.S. Department of
Agriculture of any product or service.
1
Flitcroft
Chapter 3
22
Variable Selection
A large number of instream and network variables were
available for analysis (Table 3.1). Three steps were used to
reduce this list for modeling and statistical analysis. Each
year was assessed individually because the snorkel surveys
were not designed to survey the same pools every year. First, a
correlation analysis of the original suite of variables available in
the network and habitat datasets was completed for each year
of data. Three of the network variables were excluded from the
correlation analysis (CA) because of interest in their inclusion
in later analysis and modeling (the variables described the
nearest distance to habitats important in summer, for winter
rearing and adult spawning), while one categorical variable
(valley form) was excluded from the CA of instream metrics.
Next, variables were dropped based on similarity and a
preliminary principle components analysis (PCA) (PC Ord
software, McCune and Mefford 1999). A PCA of the reduced
network and instream habitat metrics guided the final selection
of variables for models of juvenile density with instream and
network metrics. Only variables that were important in the
eigenvalues of the first three PCA axes were selected. This PCA
was also used to describe the habitat structure in each year of
data.
Objective One: Juvenile Density at the Site Scale
Juvenile coho salmon use different habitats in different
seasons during their year of freshwater residency. The
juxtaposition of these habitats in the stream network may be
important when considering patterns of juvenile occupancy
and site-specific density. In order to explore the efficacy of
network metrics in describing juvenile density, I compared
the effectiveness of instream site-specific habitat variables
with stream network variables. Data were analyzed from three
subbasins in the Alsea River basin (note: Upper Drift Creek,
Five Rivers, and the South Fork Alsea) and four subbasins in
the Siletz River basin (note: Rock, Cedar, Sams and Sunshine
Creeks) that contained overlapping juvenile snorkel survey
locations and ODFW aquatic inventory survey sites. Sites
(n=1990) included data from 1998, 1999, 2001 and 2002
juvenile surveys. Throughout the site-scale analysis, juvenile
density was log transformed to meet assumptions of normality.
Before modeling instream and network metrics, I explored
the variability within the dataset. First, an Analysis of Variance
(ANOVA) (using SAS 9.1 software) was used to test for
differences in juvenile density among and within subbasins
over the four years of data (1998, 1999, 2001 and 2002). The
distance from the site to the stream mouth was included in
the ANOVA to control for autocorrelation (tested for with
Durbin-Watson test). Second, I explored whether the habitat
used by juvenile fish changed among years. The identification
Figure 3.3. Generalized pictorial of network distance calculation:
network distances were calculated from juvenile snorkel sites to the
closest available seasonal habitat.
of predominant metrics describing habitat were explored using
a PCA of both the instream and network datasets.
In preliminary data analysis I noticed that some subbasins
had increasing juvenile density between 2001 and 2002 and
others had decreasing densities. I used a discriminant analysis
(DA) (grouping on increasing and decreasing subbasins) and
PCA to explore whether the network or habitat datasets could
help explain subbasin differences.
I next modeled juvenile density using the instream and
network datasets independently and together. I sought a
modeling technique that uses non-parametric relationships
between dependent and independent variables because
preliminary analysis using parametric relationships was
unsatisfactory due to low R2 values. I chose a local-mean
non-parametric multiplicative regression (LM-NPMR) for
modeling (Hyperniche software, McCune and Mefford 2004).
The form of non-parametric multiplicative regression used in
the Hyperniche software is characterized by a leave-one-out
cross-validation model selection technique meant to guard
against model overfitting. Model quality was evaluated using a
cross-validated R2 value and examination of model fit in graphs
of residuals and estimated values. The LM-NPMR used in my
analysis incorporates a Gaussian local mean as the local model.
The local model describes the shape of the function used to fit
each point with the variables for that model (McCune 2004).
The dependent variable for all site level analysis was the log
density of juvenile coho salmon (number of fish per meter of
stream length). This analysis was not an attempt to define the
best habitat for juveniles or adults. I recognize that density
is a misleading indicator of habitat quality because density
depends, in part, on the length of the pool in which the
fish are counted. For juveniles, large and long pools in the
lower portion of a watershed will have less unusable space for
juveniles than small and short pools in the headwaters. So
even though the larger downstream pool has a lower density,
Flitcroft
Chapter 3
it may contain higher quality habitat than the pool upstream.
In addition, juvenile coho salmon exhibit territorial behavior
so territory size may be a more accurate tool for measuring
habitat quality (Grant et al. 1998). However, in the interests of
inventorying entire watersheds in a timely manner, only counts
of fish in pools were taken by field survey crews. Describing
habitat quality based on density alone may be misleading, but
density provides a means of standardizing the data relative to
the size of the pool and is useful for models that are attempts
to understand abundance.
Objective Two: Juvenile Coho Patches
The second objective is meant to explore the physical
context and meaning of patches of juvenile coho. The spatial
autocorrelation structure of juvenile coho salmon density was
generated for each subbasin and each year using variograms.
Eleven subbasins were used in this analysis (Siletz: Cedar,
Sunshine, Rock and Sams Creeks; Alsea: Drift, Fall, Canal,
Lobster Creeks and North Fork Alsea, South Fork Alsea and
Five Rivers). In this application, the area of autocorrelation,
or variogram range, was interpreted to represent the size of
patches occupied by juvenile coho. Variograms were used that
included stream network distances between pools, rather than
Euclidean distance between pools that is the typical method.
The code and technique for using network distances were
developed by Ganio et al. (2005). Also, juvenile density was
detrended for this analysis by using the residuals of a regression
that incorporated the distance to the mouth from every site.
Distance downstream was the variable selected for detrending
because the correlation between distance downstream and
juvenile density was comparable to other network metrics and
has been used by other researchers doing similar variogram
detrending (personal communication L. Ganio, Oregon State
University).
The manner in which a stream network branches might
be important when describing patch sizes of juvenile coho
salmon. For example, a stream that is more linear with few
tributaries might reflect a different pattern of occupancy,
described by patch size, than a stream system with multiple
large branches that contained branched tributaries. Therefore,
subbasins were clustered into groups based on the average
gradient in first through third order streams. Higher-gradient
low-order streams imply a steeper geomorphology than lowergradient low-order streams. This difference in gradient could
result in different shaped basins with low gradient streams
tending toward a round and more dendritic shape and high
gradient corresponding with a linear topology. A discriminant
analysis tested whether linear and dendritic subbasins had
different patch sizes of juvenile coho salmon.
23
Objective Three: Patterns of Juvenile Occupancy in
Subbasins
Analyzing juvenile coho salmon occupancy at the subbasin
scale was meant to provide a large-scale context within
which patterns at smaller scales might better be interpreted.
Summarizing occupancy at the subbasin scale also provided
an opportunity to test whether occupancy changed between
years. Further, subbasins were compared to spawning adult
coho salmon run sizes and precipitation summarized at the
mid-coast regional scale.
Juvenile coho salmon occupancy at the subbasin scale
was summarized in a detailed count of stream kilometers
in subbasins that were surveyed in at least three of the four
survey years available. This included seven subbasins in the
Alsea River basin (note: Drift, Canal, Five Rivers, Fall and
Lobster Creeks, the North Fork Alsea and South Fork Alsea)
and four in the Siletz River basin (note: Rock Creek, Cedar
Creek, Sams Creek and Sunshine Creek). Visual inspection
has been identified as a useful preliminary assessment tool
(Torgersen et al. 2004). Subbasin maps of juvenile distribution
were examined visually to explore how juvenile occupancy
expands or contracts within the network. An unbalanced
repeated measures analysis of variance (ANOVA - using SAS
9.1 software) tested whether there was a difference in subbasin
occupancy over time. In this test, subbasins are assumed to be
independent, but subbasins themselves are correlated through
time. Therefore, subbasin forms the unit of replication.
Juvenile occupancy at the subbasin scale was explored
with respect to two-large scale environmental conditions that
might be important when exploring juvenile coho salmon
distribution: adult spawning run size, and precipitation during
spawning. In this retrospective analysis, no causal relationships
were sought. Rather, this part of the study attempts to
incorporate large-scale relationships in an interpretation of
network patterns. Adult spawning run size was chosen for
interpretation because I assumed that more parents would lead
to more juveniles. This might result in wider network use or
different patterns of juvenile density. Precipitation was chosen
for interpretation because it has been used as a surrogate
measure of streamflow (Swift et al. 1988). It is possible that in
years of higher precipitation, adults might mover further into
the network following higher water levels and more habitats.
Precipitation records for Oregon’s mid-coast were averaged
through the months of the spawning run (November through
February).
Flitcroft
Chapter 3
Results
Variable Selection
The results of the correlation analysis was the same for
each year. This could reflect the similar spatial location of sites
between years. Based on an assessment of the correlation matrix
in each year, the dataset of network variables was reduced
from eight to four (Table 3.2). The four selected variables
were stream order, elevation, mean gradient downstream, and
valley width index. Stream order was selected because it was
not highly correlated with any other network metrics. Mean
annual precipitation was removed because it was correlated
with elevation. Intrinsic potential for coho salmon, gradient
of intrinsic potential for coho salmon and total stream width
were removed because all were correlated with mean gradient
downstream. Although valley width index was also correlated
with the intrinsic potential for coho salmon, gradient of
intrinsic potential for coho salmon, and total stream width,
it was kept because it was not highly correlated with mean
gradient downstream. The correlation analysis of the sitespecific habitat variables only resulted in the exclusion of
wood volume, because of its high correlation with the wood
count (Table 3.2).
Although the correlation analysis was useful in reducing
variables in the network dataset, it did not make a significant
reduction in the number of variables in the site-specific
habitat dataset. However, I opted to use five of the seventeen
site-specific habitat variables to represent the entire dataset.
This selection was guided by a PCA and also by the biological
Table 3.2. Summary of correlation matrix showing significant
correlation relationships (> 0.60) used in variable reduction from
Oregon’s Mid-Coast region in 1999 (n = 987), 2001 (n = 1288),
and 2002 (n = 1416): Habitat and network datasets.
Selected Variable
Network Dataset
Elevation
Correlated Variable
Mean annual
precipitation
Mean gradient
Intrinsic potential for
downstream
coho salmon
Gradient of potential
for coho salmon
Valley width index Total stream width
Intrinsic potential for
coho salmon
Gradient of potential
for coho salmon
Total stream width
In-Stream Habitat
Wood Count
Wood Volume
Correlation Values
by year:
1999 2001 2002
0.81
0.72
0.69
-0.77 -0.76 -0.76
-0.74 -0.74 -0.74
-0.77 -0.69 -0.71
0.74 0.69 0.68
0.69
0.64
0.64
0.72
0.65
0.66
0.83
0.79
0.71
24
importance of certain physical features. This variable reduction
was an attempt to avoid duplication of metrics. The five
representative habitat variables are valley form (a categorical
variable), water depth, percent sand, percent gravel, percent
undercut bank, and number of pieces of wood. Valley form
was meant to encapsulate slope, shade and canopy closure.
Percent sand and gravel represented substrate, while silt/
organics, cobble, boulder, bedrock and boulder count were
omitted. Also, percent sand measured fine-grained material,
which can be detrimental to juvenile survival to emergence,
and gravel measured the coarse-grained substrate necessary for
spawning adults (Groot and Margolis 1991). Undercut bank
was chosen as an indicator of active channel erosion. Number
of pieces of wood was chosen to represent all wood variables,
including wood volume and key pieces of wood.
To further reduce the number of instream habitat variables
for modeling purposes, an additional level of variable selection
incorporated a principle components analysis for each year
of data. The results were different for every year (Table 3.3
for network dataset and Table 3.4 for instream habitat), so
any variable that was identified throughout the four years was
included in the final list of network and instream variables to be
used for modeling. These variables were chosen for modeling
because they were important in describing the variation in
habitat for these two datasets over the four years of study data
(network variables: distance to summer pools, distance to
spawning habitat, distance to rearing habitat, order, subbasin,
valley width index and mean gradient downstream; in stream
variables: valley form, percent gravel, percent sand, undercut
and number of pieces of wood).
Objective One: Site Scale Patterns
Objective one incorporated stream network metrics into an
analysis at the site scale. This process explored the usefulness
of stream network metrics and how they contribute to an
understanding of habitat use by juvenile fish in the context
of their diverse life history (Groot and Margolis 1991) and
behaviors (Chapman 1962, Nielsen 1992). Before exploring
network metrics directly, I tested the subbasins to see if juvenile
density or habitat configurations changed across years.
Analysis of Variance (ANOVA) results indicated that there
was a subbasin-by-year effect in the densities (p-value < 0.001)
(Table 3.5). This means that juvenile density changes within
subbasins in different ways across years. The ratio of change
between these years varied by subbasin with some subbasins
showing lower densities between years (ratio > 1.0) and some
subbasins with higher densities (ratio < 1.0) (Table 3.6).
Sunshine Creek was the only subbasin in which 1998 had
a higher density relative to later survey years (1998 to 1999
ratio = 24.6132, 1998 to 2001 ratio = 7.3161, and 1998 to
2002 ratio = 3.8349). Cedar Creek was the only subbasin
Flitcroft
Chapter 3
25
Cumulative
Percent of Broken-stick Dominant Variables in
Variance
Eigenvalue Eigenvector
Table 3.3. Principle components
Percent of
analysis results for sites in Oregon’s
Year Axis Eigenvalue Variance
Mid-Coast in 1998, 1999, 2001,
and 2002, using a dataset of variables 1998 1
2.342
33.451
describing network relationships.
2
1.51
21.575
3
1.078
15.407
2.593
55.026
70.433
1.593
1.093
Distance to summer pools;
Distance to spawning
Order
Subbasin
1999
1
2
3
1.61
1.36
1.31
26.88
22.62
21.81
26.88
49.50
71.31
2.45
1.45
0.95
Order; Valley width index
Mean grad downstream
Distance to rearing
2001
1
2.06
34.26
34.26
2.45
2
3
1.46
1.09
24.39
18.22
58.65
76.78
1.45
0.95
Distance to spawning;
Distance to rearing
Mean grad downstream
Valley width index
1
2.11
35.20
35.20
2.45
2
3
1.48
1.11
24.73
18.41
59.93
78.34
1.45
0.95
2002
Table 3.4. Principle components
analysis results for sites in Oregon’s
Mid-Coast in 1998, 1999, 2001,
and 2002, using a dataset of
variables
describing
instream
habitat conditions.
33.451
Year
Distance to spawning;
Distance to rearing
Mean grad downstream
Valley width index
Cumulative
Percent of Broken-stick Dominant Variables in
Variance
Eigenvalue Eigenvector
Percent of
Axis Eigenvalue Variance
1998
1
2
3
1.854
1.292
1.121
26.491
18.463
16.019
26.491
44.954
60.973
2.593
1.593
1.093
Valley form
Percent gravel
Undercut
1999
1
2
3
2.07
1.23
1.01
29.50
17.62
14.44
29.50
47.12
61.55
2.59
1.59
1.09
Percent sand; Percent gravel
Undercut
Number wood pieces
2001
1
2
3
1.99
1.24
1.06
28.40
17.74
15.10
28.40
46.14
61.24
2.59
1.59
1.09
Percent sand
Valley form
Number wood pieces
2002
1
2
3
2.14
1.26
1.12
30.60
17.93
15.96
30.60
48.53
64.50
2.59
1.59
1.09
Percent sand
Valley form
Number wood pieces
Table 3.5. ANOVA results examining juvenile coho density and subbasins in Oregon’s Mid-Coast region over time.
Log Juvenile Coho Density ~
Year + subbasin + year*subbasin + downdist
Effect
df
F Value
Pr > F
Year
Subbasin
Year*subbasin
Downdist
3
6
13
1
35.82
13.2
18.8
15.15
< 0.0001
< 0.0001
< 0.0001
0.0001
Test of whether there were significant differences in juvenile coho
density within individual subbasins over available years.
Subbasin
df
F value
Pr > F
Cedar
Upper Drift
Five Rivers
Rock
Sams
S.F. Alsea
Sunshine
2
2
2
3
2
2
3
18.05
35.99
6.85
54.08
6.46
26.39
21.96
< 0.0001
< 0.0001
0.0011
< 0.0001
0.0016
< 0.0001
< 0.0001
Flitcroft
Chapter 3
26
Table 3.6. Pairwise comparisons of the difference in juvenile coho density between years for each subbasin.
Start Year
1998
1998
1998
1998
1998
1998
1998
1998
1998
1998
1999
1999
1999
1999
1999
1999
1999
1999
1999
1999
2001
2001
2001
2001
2001
2001
2001
Start Subbasin To Year To Subbasin
Five Rivers
Five Rivers
Rock
Rock
Rock
S.F. Alsea
S.F. Alsea
Sunshine
Sunshine
Sunshine
Cedar
Cedar
Upper Drift
Upper Drift
Rock
Rock
Sams
Sams
Sunshine
Sunshine
Cedar
Upper Drift
Five Rivers
Rock
Sams
S.F. Alsea
Sunshine
2001
2002
1999
2001
2002
2001
2002
1999
2001
2002
2001
2002
2001
2002
2001
2002
2001
2002
2001
2002
2002
2002
2002
2002
2002
2002
2002
Five Rivers
Five Rivers
Rock
Rock
Rock
S.F. Alsea
S.F. Alsea
Sunshine
Sunshine
Sunshine
Cedar
Cedar
Upper Drift
Upper Drift
Rock
Rock
Sams
Sams
Sunshine
Sunshine
Cedar
Upper Drift
Five Rivers
Rock
Sams
S.F. Alsea
Sunshine
bt lower
Ratio
br upper
0.0738
0.1320
0.2759
0.0020
0.0036
0.0283
0.1203
11.0252
3.5005
1.8910
2.9093
0.1764
0.0213
0.0629
0.0074
0.0054
0.1220
0.1874
0.1310
0.0706
0.0223
1.3429
0.7740
0.2757
0.7384
1.8075
0.2542
0.1829
0.3233
0.6882
0.0129
0.0094
0.0608
0.2345
24.6132
7.3161
3.8349
13.8146
0.8007
0.0448
0.1281
0.0188
0.0137
0.2587
0.3944
0.2972
0.1558
0.0580
2.8559
1.7698
0.7265
1.5245
3.8551
0.5242
0.4523
0.7913
1.7169
0.0334
0.0243
0.1308
0.4574
54.9477
15.2905
7.7646
65.5982
3.6336
0.0870
0.2609
0.0480
0.0349
0.5486
0.8398
0.6744
0.3436
0.1503
6.0864
4.0467
1.9161
3.1474
8.2221
1.0808
with a higher density of juvenile coho in 1999 than in 2001
(ratio of 13.8146). With the afore mentioned exceptions of
Sunshine and Cedar Creeks, all subbasins had higher density
between years from 1998 through 2001. However, between
2001 and 2002 several subbasins had increasing densities of
juvenile coho salmon while others had decreasing densities.
Drift Creek, Five Rivers, Sam Creek and the South Fork Alsea
all had higher densities in 2001 than in 2002. Cedar, Rock
and Sunshine Creeks had higher densities in 2002 than 2001
(Table 3.6).
The principle components analysis indicated that the
variation in the habitat characteristics was portioned similarly
each year. In 1998 and 1999 the set of habitat variables that
dominate axes 1, 2 and 3 are slightly different in both the
network and instream datasets, with 2001 and 2002 appearing
nearly identical (Tables 3.3 and 3.4). The network variables
of distance to summer pools, distance to spawning habitat,
distance to rearing habitat, order, subbasin, valley width index
and mean gradient downstream dominated one of the first
three pca axes in at least one year of data. From the instream
dataset, valley form, percent sand, percent gravel, undercut
and the number of pieces of wood dominated one of the first
three pca axes in at least one year of data.
The explanatory ability of the network and instream datasets,
independently and together, in describing the abundance of
juvenile coho salmon was explored using a local-mean, nonparametric multiplicative regression (LM-NPMR) process.
The best model for each dataset was selected by a comparison
of R2 values and model fit in graphs of residuals and estimated
values. In every year the best network models (1998, R2 =
0.3624; 1999, R2 = 0.1716; 2001, R2 = 0.2571; 2002, R2 =
0.2433) had higher explanatory power than the best instream
models (1998, R2 = 0.0594; 1999, R2 = 0.1178; 2001, R2 =
0.1510; 2002, R2 = 0.1050) for the log of juvenile density.
Models from the combined dataset in each year had a slightly
higher explanatory power than the network models (1998, R2
= 0.3896; 1999, R2 = 0.2345; 2001, R2 =0.2873; 2002, R2
= 0.2861). Because the network and combined models have
such similar R2 values neither sets of models may be deemed
better than the other.
Variables selected in LM-NPMR varied for each year and
dataset (Table 3.7). In the network dataset, the variables
subbasin, distance to rearing habitat, distance to spawning
habitat, distance to summer pools, mean gradient downstream
and stream order were present at least once. In the instream
habitat dataset, the variables percent sand, percent gravel,
Flitcroft
Chapter 3
wood pieces, valley code, and undercut were present at least
once.
A discriminant analysis (DA) was conducted to test
whether the observation of increasing or decreasing density
ratios within subbasins between 2001 and 2002 could be tied
to stream network or instream characteristics. The DA found
a statistically significant division between these two clusters in
both datasets (Hotelling’s T-Squared value of < 0.001 for both
datasets) (Table 3.8). The log of juvenile density, coded for
increasing or decreasing subbasins, was graphically overlaid on
the network and instream PCA ordinations for 2002 (a PCA
27
for 2002 was used because the PCAs for 2001 and 2002 were
nearly identical). There appeared to be little differentiation
between subbasins with increasing or decreasing juvenile
density in the instream dataset (Figure 3.4). However, there
appeared to be a demarcation between these two clusters of
subbasins in the network dataset along the first axis, which
predominately represents the juxtaposition variables of
distance to spawning and distance to rearing (Figure 3.5).
This is visible in the graph as the clumping of grey and black
triangles (sites) into two groups along the gradient of the first
axis.
Table 3.7. Local-mean non-parametric multiplicative regression model results for 1998, 1999, 2001, and 2002 for sites in Oregon’s MidCoast region. The dependent variable is log juvenile coho density.
Year
Variable set
Variable 2
Variable 3
Variable 4
Order
0.0594 Valley code
0.3896 Subbasin
Distance to
rearing habitat
Percent sand
Wood pieces
Mean gradient
downstream
Wood pieces
Mean gradient
downstream
1999 Network Variables
Instream Variables
Instream and Network Variables
0.1716 Order
0.1178 Percent sand
0.2345 Distance to
rearing habitat
Valley width index Subbasin
Percent gravel
Undercut
Mean gradient
Valley code
downstream
2001 Network Variables
Distance to
summer pools
Undercut
—
Instream Variables
0.2571 Distance to
Mean gradient
spawning habitat downstream
0.151 Valley code
Percent sand
Instream and Network Variables
0.2873 Percent sand
Distance to
spawning habitat
Mean gradient
downstream
Distance to
summer pools
0.2433 Distance to
rearing habitat
0.105 Valley code
0.2861 Valley code
Mean gradient
downstream
Percent sand
Undercut
Distance to
Subbasin
summer pools
Percent gravel
Undercut
Distance to rearing Subbasin
habitat
1998 Network Variables
Instream Variables
Instream and Network Variables
2002 Network Variables
Instream Variables
Instream and Network Variables
R2
Variable 1
0.3624 Subbasin
Figure 3.4. Principle Components Analysis
graph of 2002 habitat dataset displaying the
number of pieces of wood with subbasins
of increasing or decreasing juvenile density
(between 2001 and 2002) as an overlay. Each
triangle corresponds with a specific pool and
the size of the triangle represents the relative
number of pieces of wood associated with each
pool. The x-axis is associated a gradient of greater
to smaller percentages of fine substrate material
and the y-axis is associated with smaller to
greater percentages of coarse substrate material.
Percent gravel
Order
—
Wood pieces
Subbasin
Wood Pieces
Flitcroft
Chapter 3
28
Figure 3.5. Principle Components Analysis graph
of 2002 network dataset displaying distance to
spawning habitat with basins of increasing or
decreasing juvenile density (between 2001 and
2002) as an overlay. Each triangle corresponds
with a specific pool and the size of the triangle
represents the measured distance to spawning
habitat associated with each pool. The x-axis is
associated with a gradient of smaller to larger
distances between seasonal habitats while the
y-axis represents a continuum of mean gradient
downstream.
Objective Two: Patches and Basin Shape
Variograms were generated in order to compare the spatial
clumping pattern exhibited by the log of juvenile coho salmon
density (detrended with the variable distance downstream) in
the stream network among subbasins over time (Appendix 3.2).
The robust variograms allow for the estimation of the interval
(estimated variogram range) in which survey points exhibit
spatial autocorrelation. Because the range was estimated, it
included an unquantifiable amount of error. Therefore, no
rigorous statistical tests were completed using this metric as a
dependent variable. However, this metric may be summarized
graphically. Examining the graph of variogram range size per
year in each subbasin indicated that range tends to change
within subbasins among years (Figure 3.6). In addition, the
estimated variogram range size appears to be smaller in small
subbasins compared to larger subbasins. Variogram shape also
Figure 3.6. Juvenile coho salmon
patch size (estimated variogram range)
for 1998, 1999, 2001 and 2002 in
subbasins of the Alsea and Siletz River
basins. The subbasins are listed from
smallest to largest.
appeared to change as watershed area increased with a small
“hump” appearing low in the variogram of larger watersheds
(Figure 3.7). This “hump” pattern is present in at least one
year for six of the eleven subbasins in this analysis (North Fork
Alsea, Upper Drift, Rock and Lobster Creeks, South Fork
Alsea and Five (Figure 3.8). The shapes of the subbasins in the
cluster with lower average gradients appeared more rounded
and branched than the subbasins with higher average gradients.
Therefore, two cluster groups of branched and linear subbasins
were identified. Discriminant analysis of the two basin shape
clusters for 1998, 1999, 2001 and 2002 incorporated network
summary information such as area, stream kilometers occupied
by juvenile coho, number of streams occupied by juvenile
coho, total juvenile coho counted and estimated variogram
range size. None of the DA resulted in statistically significant
differentiation between basin shape clusters (Table 3.8).
Flitcroft
Chapter 3
29
Figure 3.7. Variograms for subbasins in Oregon’s Mid-Coast (in each year available between 1998 and 2002) with arrows pointing to
“hump” shape that is common in larger basins and may be associated with a nested spatial autocorrelation structure. Lines show the upper
and lower bounds of a 95% confidence interval.
Flitcroft
Figure 3.7. Continued.
Chapter 3
30
Flitcroft
Figure 3.7. Continued.
Chapter 3
31
Flitcroft
Chapter 3
32
Figure 3.8. Dendrogram showing two clusters of
subbasins described as branched or linear in shape
due to their geomorphology.
Table 3.8. Discriminant Analysis results for two sets of analysis of groups in subbasins of Oregon’s Mid-Coast: groups of subbasins
identified as increasing or decreasing juvenile coho density (between 2001 and 2002); groups of subbasins classified as linear or branched
in shape.
Hotelling’s
T- Squared
Group
Variable
Category Variable List
<0.001
Increase or Decrease Habitat
<0.001
Increase or Decrease Network Distance to spawning; Distance to rearing; Distance to summer pools;
Order; Mean gradient downstream; Valley width index
0.53
Linear or Branched 1998
Area; Stream km occupied by juvenile coho; Number of streams
occupied by juvenile coho; Juvenile coho count; Range size
sample size too small Linear or Branched 1999
Area; Stream km occupied by juvenile coho; Number of streams
occupied by juvenile coho; Juvenile coho count; Range size
0.82
Linear or Branched 2001
Area; Stream km occupied by juvenile coho; Number of streams
occupied by juvenile coho; Juvenile coho count; Range size
0.7
Linear or Branched 2002
Area; Stream km occupied by juvenile coho; Number of streams
occupied by juvenile coho; Juvenile coho count; Range size
Percent sand; Percent gravel; Undercut; Wood pieces; Valley code
Objective Three: Subbasins and Juvenile Network
Occupancy
Stream occupancy descriptors such as stream kilometers
occupied and number of streams occupied per year were
summarized for each subbasin (Table 3.9). An unbalanced
repeated measures ANOVA was used to test whether subbasin
occupancy changed similarly over time within and among
subbasins. There were statistically significant differences in
juvenile coho salmon occupancy among years (p < 0.001, df
= 3, F value = 9.19) and also among subbasins within years (p
< 0.001 with Bonferroni correction of alpha level to 0.0125)
(Table 3.10). With a Tukey-Kramer adjustment, differences in
subbasin occupancy between years were statistically significant
(p-value < 0.05) for three pairs of years (1998 to 2002 adjusted
p = 0.0169; 1999 to 2001 adjusted p = 0.0081; 1999 to 2002
adjusted p = 0.0021) (Table 3.11). After Torgersen et al.
(2004), a visual inspection of subbasin maps (Appendix 3.3)
was completed to suggest further analysis or interpretation.
The patterns of juvenile coho salmon distribution in the maps
of network occupancy imply that when higher numbers of
stream kilometers are occupied, the distribution of juvenile
coho salmon moves into the headwaters, while lower numbers
of stream kilometers occupied has the reverse effect (Figure
3.9).
The patterns of juvenile coho salmon stream occupancy in
the subbasins of the Alsea and Siletz appear similar to pattern
of the adult spawning run size for the entire Mid-Coast region.
Adult spawning run sizes increase between 1997 (parents
of 1998 juveniles) and 2001 (parents of the 2002 juveniles)
(Figure 3.10). Subbasin occupancy summary information
shows an increase in the number of stream kilometers
occupied during this time period in most basins. Fall, Lobster,
Sams, and Sunshine Creeks showed increases and decreases
in occupancy among years rather than a consistent increase
Flitcroft
Chapter 3
Table 3.9.
Summary table
showing number of sites
occupied, total number of sites
surveyed, juvenile coho count,
number of streams occupied and
stream miles occupied for the 11
subbasins of interest in Oregon’s
Mid-Coast region.
Alsea River Basin
Year
Canal Creek basin
(32.964 km2)
1998
2001
2002
Fall Creek
1998
(71.80 km2)
2001
Hatchery in mid-basin
2002
NF Alsea
1998
(73.532 km2)
2001
Below hatchery
2002
Upper Drift Creek
1999
(79.66 km2)
2001
2002
Lobster Creek
1998
(143.71 km2)
1999
2001
2002
SF Alsea
1998
(158.79 km2)
2001
2002
Five Rivers (299.95 km2)
1998
1999 data excluded uneven 2001
coverage with other years
2002
33
# Sites
Surveyed
# Sites
Occupied
# Juvenile
Coho
# Streams Stream (km)
Occupied Occupied
90
135
156
127
108
107
186
195
277
328
267
308
476
389
411
487
230
189
268
648
632
922
88
85
130
86
63
58
112
135
198
231
234
308
313
257
351
431
107
138
216
474
501
756
1082
2753
2588
1134
2113
1057
1461
3483
4607
2990
19162
10919
4010
2499
12033
13279
350
1825
3086
5332
18808
18782
4
8
9
7
5
6
6
11
15
16
13
15
17
18
23
22
7
7
10
25
33
48
11.806
14.244
18.121
17.620
16.234
17.185
19.125
24.793
29.027
28.351
30.728
40.446
41.488
29.788
60.934
65.795
94.555
107.317
115.741
62.596
81.465
97.460
16
83
121
283
233
303
138
85
173
175
239
239
267
295
10
67
81
202
200
269
102
54
133
134
103
93
184
224
84
4074
2029
2194
8826
8841
2352
470
5515
6360
601
789
6378
3902
1
4
5
10
10
10
6
4
9
7
8
8
13
10
4.328
14.245
17.551
15.708
14.678
18.956
13.673
7.896
16.423
13.13
21.448
28.442
34.852
38.999
Siletz River Basin
CedarCreek
(33.44 km2)
Sams Creek
(37.71 km2)
Sunshine Creek
(77.0 km2)
Rock Creek
(104.72 km2)
Table 3.10. Unbalanced repeated measures ANOVA for
subbasins in the Alsea and Siletz Basins that tested for
differences in least squares means of the number of stream
kilometers occupied by juvenile coho between 1998 and 2002.
Year
1998
1999
2001
2002
Juvenile Coho
Salmon Mean
Occupancy Standard
(km)
Error
30.76
27.83
37.86
43.00
9.43
9.53
9.35
9.35
Confidence
Interval
11.20 - 50.32
8.06 - 47.60
18.48 - 57.24
23.62 - 62.38
1999
2001
2002
1999
2001
2002
1998
1999
2001
2002
1998
1999
2001
2002
Table 3.11. Unbalanced repeated measures ANOVA for subbasins in
the Alsea and Siletz Basins that tested for differences in least squares
means in the number of stream kilometers occupied by juvenile coho
salmon between years from 1998 and 2002.
Juvenile Coho
Salmon Mean
From
Occupancy
Year To Year
(km)
1998
1998
1998
1999
1999
2001
1999
2001
2002
2001
2002
2002
2.9314
-7.0982
-12.233
-10.03
-15.164
-5.1346
Standard
Error
3.2622
2.941
3.7361
2.7881
3.6282
2.3522
TukeyKramer
df t-value Adjusted P
22
22
22
22
22
22
0.9
-2.41
-3.27
-3.6
-4.18
-2.18
0.8056
0.1038
0.0169
0.0081
0.0021
0.1592
Flitcroft
Chapter 3
34
Figure 3.9. The series of Canal Creek maps (1a, 1c, 1d) show an increase in juvenile network occupancy upstream from 1998 to 2002 while
Sunshine Creek (2a-2d) shows variation in network occupancy including a decrease in occupancy between 1998 and 1999 and 2001 and
2002.
1a. Canal Creek 1998
##
#
#
##
###
# ##
##
S
##
S
##
S
#
##
#
S
#
S
#
S
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
S#
#
S
#
#
##
#
#
#
S
#
#
#
#
#
#
#
#
##
S
#
S
S
#
S#
#
S
S
#
##
S
#
##### #
S
#
#
S
#
S
#
S
#
S
#
S
#
S ##
S#
S
#
####
S
#
S
#
S#
S#
S#
#
#
S
#
S
#
#
#
S
#
#
#
#
#
#
#
#
#
S
S
S
#
1c. Canal Creek 2001
#
#
#
#
#
# ##
#
#
#
##
#
S
#
#
#
#
#
#
#
#
#
# #
### #
S
#
S
#
S
#
S
#
S#
#
S#
S
#
#
##
S####
S
S#
#
#
S#
#
S#
#
##
S#
#
##
#
#
S
#
S
#
S
#
S
#
S#
#
#
S#
S#
S#
S#
##
S
S
#
#
#
#
S
#
#
###
S
#
#
#
S
#
#
S
#
#
#
##
S
S#
#
#
#
S#####
#
#
#
S###
S#
S#
#
S
#
S#
#
###
#
##
S
(no data for Canal Creek 1999)
Present
Absent
1d. Canal Creek 2002
##
#
###
#
##
##
#
##
#
S
#
#
#
#
#
##### ###
####
#
#
#
#
##
##
##
S#
S#
#
#
##
#
##
# ####
##
##
#
#
#
#
#
#
##
S## #
S#
S# #
#
S#
#
#
#
S
#
S###
#
#
#
#
#
S
#
#
#
S
#
#
#
#
#
S
#
#
#
#
# #
S#
###
S
#
#
#
#
#
#
##
S
S
#
#
###
S
#
S
#
#
#
#
#
#
###
###
S
#
#
S
#
S ###
S#
#
#
#
##
S
#
Flitcroft
Chapter 3
35
Figure 3.9 (continued).
2a. Sunshine Creek 1998
S
#
S
#
S
###
#
#
##
#
#
S
#
#
S
#
S
#
##
S
#
S
#
S#
S#
#
S
####
#
S
#
#
#
#
#
#
#
#
#
S
#
#
S
#
S
#
#
## # ## #
#
##
S
#
##
#
##
S##
#
S ## ##
#
##
#
###
#
#
#
##
##
##
#
#
#
##
#
S
#
#
S
#
#####
##
#
#
#
##
S#
##
##
###
#
Present
###
S
#
2b. Sunshine Creek 1999
S#
###
S
#
S
#
S
#
S
####
S#
#
S
##
#
#
#
#
S
#
#
S
S#
#
#
S#
S ##
S#
##
S
#
#
#
##
S
#
#
#
#
#### #
##
# ##
##
S# #
S
#
S
#
S#
#
S
#
S
#
### ##
##
#S
## #
S
S
###
Absent
2c. Sunshine Creek 2001
#
###
#
#
##
#
S
#
#
##
S#
#
#
#
##
##
##
#
S
#
#
S
#
#
##
#
#
#
#
#
#
#
#
S
#
#
S
#
#
#
## # # #
#
#
#
##
#
S # #
#
#
S
#
##
### ##
S
#
S
#
#
#
#
#
###
#
#
#
#
#
#
#
#
#S
##
S
#
S
#
#
#
##
#
#
#
#
#
#
### #### ##
#
#
S
#
S
#
####
#####
##
#
#
#
##
#
#######
## #
SS
S#
##
Figure 3.10. Number of coho salmon spawners in
Oregon’s Mid-Coast and precipitation during the
spawning run (October through March).
2d. Sunshine Creek 2002
#
#
#
#
#
#
#
#
#
S#
##
S
#
#####
#####
#
#
#
#
#
#
#
#
S
#
#
S
#
S
#
#
#
#
#
## # ## ##
#
S
#
S
#
#
S
#
S
#
#
#
S
#
#
S
#
#
##
S
## #
#
S#
###
#
#
S
S#
S#
#
###
##
#
#
#
S
#
##
S
#
S
#
#
#
#
##
#
#
##
#
S# # ### #
#
#
#
#
#
#
#
#
##
#
##
##
#
#
##
#
#
#
#
##
#
#
#
##
#
##
#
S#
#
S
Flitcroft
Chapter 3
(Table 3.9). Also, in all basins except Sunshine Creek, the
number of kilometers occupied by juveniles increased between
2001 and 2002. Precipitation during the spawning run time
for the entire mid-coast was the highest in 1998 and lowest in
2000 (Figure 3.10).
Discussion
Patterns of juvenile coho abundance, distribution
and occupancy were analyzed at three spatial scales that
incorporated elements of stream network structure in variables
or statistical analyses. The analysis and synthesis of these scales
provides an opportunity to consider the complex population
ecology of coho salmon.
Objective One: Site Scale
Site-scale analysis allows for the exploration of local
habitat conditions that may be important for individual
fish. Results show that juvenile density changes among years
within subbasins, but the core complement of stream habitats
occupied by juvenile coho salmon remains similar over time.
This means that while juvenile density may change over time,
the fish tend to occupy the same types of habitat presumably
in the same portions of the stream network. Juveniles appear
to have high site fidelity (Bell et al. 2001) and occupancy in
similar habitats in the network is consistent with the tendency
of the majority of adults to home to their natal habitats to
spawn (Labelle 1992). Site scale analysis highlighted the
relevance of a stream network framework in exploring juvenile
coho salmon density. Although the overall explanatory power
of models using the instream or network datasets individually
or together was low, even for ecological datasets, network
variables were consistently useful in each year. Clearly, other
variables that were not captured in this analysis explain most
of the variation in the juvenile coho salmon dataset. These
might include micro habitat variables, intra and interspecies interactions, food availability and predation. The low
explanatory power of the models could also be due to the
ability of juveniles to exploit a wide variety of pool habitats,
as is consistent with the diversity of their behaviors (Nielsen
1992). However, it is important to note that the spatial
arrangement or topology of stream habitats has been shown
to be helpful in assessments of habitat quality in other species
(Jones et al. 2003) and explains the density patterns of juvenile
coho salmon at least as well as instream habitat metrics.
Juvenile coho salmon density might have been expected
to increase or decrease in synchrony across subbasins as
each subbasin responded to the same set of environmental
conditions such as rainfall. Indeed, juvenile density appeared
to increase or decrease relatively consistently across years
within subbasins except between 2001 and 2002. The lack of
36
consistency in changes of juvenile coho salmon density among
subbasins between 2001 and 2002 is intriguing and led to an
investigation that pointed to network variables, in particular
habitat juxtaposition variables, as a possible difference
between subbasins having opposite variation in juvenile
density between 2001 and 2002. This study’s results imply
that in some years, juvenile density is higher in watersheds
with close proximity among important seasonal habitats. This
interpretation is supported by Kocik and Ferreri (1998) who
found that the productivity for Atlantic salmon (Salmo salar)
depends on the proximity of the array of habitats necessary for
freshwater spawning and rearing.
Dynamic stream conditions include disturbance, climate
and physical habitat, and mean that different sets of habitats or
entirely different subbasins may be most hospitable for juvenile
coho salmon over time (Reeves et al. 1995, Quinn and Peterson
1996). I suspect a difference in a large scale environmental
condition that may have made subbasins with a large distance
between seasonal habitats a greater impediment to juvenile
survival in 2002 compared to 2001. One possible difference
could be related to streamflow during emergence. Stream
flow levels have been shown to effect juvenile abundance in
other stream fishes (Schlosser 1985). For salmonids, higher
flow could possibly scour redds (Montgomery et al. 1999)
reducing survival to emergence. Also, high streamflow could
impair movement of juvenile coho that are less adapted to
high velocity environments than other salmonids (Bisson et
al. 1988). Therefore, moving from spawning beds to summer
rearing pools could be more difficult under high flow conditions
in subbasins in which the proximity between seasonal habitats
is great. Unfortunately, precipitation and streamflow measures
which would be necessary to test this hypothesis were not
available for individual subbasins. Another, simpler reason for
the differences in juvenile numbers between 2001 and 2002
could be that lower numbers of spawning adults returned to
some of the subbasins thereby creating fewer progeny to be
counted in the following summer. Unfortunately, detailed
spawner estimates at the subbasin scale that track spawner
occupancy patterns are not available. Further research that
tracks adult spawner counts, streamflow, and precipitation at a
subbasin scale is necessary to fully understand the implications
of habitat topology and stream conditions on juvenile coho
salmon density and survival.
Objective Two: Patch Scale
Salmonids, like other species, are not randomly distributed
across the landscape (Dunham et al. 2002). Coho salmon are
no exception and individuals are found in patches that may
correspond with behavior, intra and interspecies interactions,
predation or habitat. Habitat is often associated with
environmental gradients such as elevation or a hierarchy of
Flitcroft
Chapter 3
nested organization such as stream order. Understanding how
and why individuals are distributed on the landscape may
contribute to ecological knowledge about habitat preferences
and survival thresholds across spatial scales (Peters et al. 2006).
The size of the area within which measures of juvenile coho
salmon are correlated might help explain the spatial structure
of the stream network as it is interpreted by juvenile fish. This
structure in turn may inform interpretations of population
dynamics at multiple spatial scales.
The spatial extent of autocorrelation within the stream
network was measured using variograms of juvenile coho
salmon density, controlling for spatial network position.
Results indicate that there are differences in estimated
variogram range, or patch size, among years both within and
across subbasins. It also appears that subbasin area may be
related to patch size. While it is intuitive that smaller subbasins
might have smaller clusters of juvenile coho salmon, such an
observation has not been quantified before. Small subbasins
tended to have smaller patch sizes than larger subbasins, and
larger subbasins exhibited symptoms of nesting. Patch size
could represent the neighborhood of habitats accessible to a
group of interacting juvenile coho salmon. Therefore, patches
may correspond to the predominant range of movement for
juvenile coho salmon between available habitats in a particular
stream system. The similarity and variation in patch sizes
identified in this study may support the definition of a patch as
a grouping of juvenile coho salmon around available habitats
within the stream network.
In site-scale analysis, the topology or orientation of habitats
appeared to be an important distinction among subbasins
when exploring changes in juvenile density. It would be
reasonable to assume that more habitats are available in a
larger stream network and that juveniles may move further
among habitats in larger systems. This could correspond to
larger range sizes in bigger subbasins, as was found in this
study. However, other research indicates that juvenile coho
salmon tend to access locally available habitats within their
natal stream systems. Bell et al. (2001) showed that juvenile
coho salmon exhibit high site fidelity in off-channel habitat
during winter floods. Also, Lonzarich et al. (2000) found that
in pool- riffle habitat sequences, fish will move out of pools
adjacent to short riffles more than from pools adjacent to
long riffles. Therefore, the hierarchical organization and larger
patch sizes in larger subbasins could correspond to overlapping
patches of juvenile coho salmon rather than a cohesive group
of fish and habitats. Further exploration of patch size and area
is necessary to more fully describe and identify the ecological
significance of variogram range sizes.
I speculated that basin shape could influence the pattern
of juvenile coho occupancy reflected by patch size. Basin
shape was explored by clustering watersheds based on the
average gradient of low-order streams to identify two distinct
37
groups. The shape of the watersheds with low gradient
headwaters appeared to be rounder (branched) than the
shape of watersheds with higher-gradient headwaters (linear).
However, the patch sizes of juvenile coho that were identified
using variograms were not different between the two shape
groupings. It is possible that this dataset did not contain
enough variation in geology or geomorphology to detect
differences in juvenile coho salmon patch size between basin
shape clusters. Unpublished research conducted by Christian
Torgersen (2006 personal communication, USGS, Seattle,
Washington) links patch sizes of coastal cutthroat trout (O.
clarki clarki) with underlying geology across a large region.
Further research across a larger regional frame incorporating
different underlying geologic structure is necessary to more
fully explore the question of watershed shape and patch sizes
of juvenile coho salmon.
What can be deduced from the results of the patch
analysis is that juvenile coho salmon patch sizes change across
years, and variogram shapes suggest that larger watersheds
include multiple scales of patchiness in juvenile coho salmon
abundance. Such differences in spatial arrangement between
small and large watersheds imply the need for different
management expectations based on watershed size. While
larger watersheds may indeed produce more fish, the nested
structure of spatial autocorrelation could mean that the
relationship between watershed size and fish numbers is
not linear. This implies that comparisons of patch and fish
population sizes should be done among basins of similar size.
Objective Three: Subbasin Scale
The convergence of life history stages in patterns of juvenile
occupancy informs an interpretation of population processes at
the subbasin scale. The number of stream kilometers occupied
by juvenile coho salmon at the subbasin scale increased in 7 of
11 subbasins between 1998 (or 1999) to 2002. This parallels
the trend of increasing mid-coast spawning adult runs during
these years. However, some subbasins did not show steadily
increasing numbers of stream kilometers occupied as the
spawning run size increased. Because accurate counts of
spawning run sizes and distributions at the subbasin scale are
unavailable, I cannot discern whether some subbasins simply
had spawning run sizes that did not match the overall midcoast trend. Therefore, greater variability in run sizes that the
subbasin scale could correspond with the lower numbers of
kilometers occupied by juveniles that were observed in other
basins. What is discernable at the subbasin scale is that there
is greater variation within and among subbasins over time in
the amount of stream kilometers occupied by juvenile coho
salmon than is found in the overall size of the adult spawning
run. Increasing variation coupled with smaller spatial extents
has been established in other studies (i.e., Wimberly et
Flitcroft
Chapter 3
al. 2000). Further research that includes the collection of
accurate estimates of precipitation, adult spawning run size,
and streamflow at the subbasin scale would allow for greater
discernment of the interrelationship between life stages that
may be interpreted from patterns of stream network occupancy
by juveniles.
Maps of individual subbasins showed that when stream
network occupancy was high, juvenile coho were found
upstream in the headwaters, while in years of low stream
network occupancy juvenile coho salmon were only lower
in the subbasin. In seven of the eleven subbasins assessed,
occupancy by juvenile coho salmon seemed to increase and
decrease in relation with adult spawner run sizes in the entire
mid-coast over the 5 years of this study. It appears then, that
in these subbasins, it is possible that more spawning adults
resulted in a greater portion of the stream network occupied
by juvenile coho salmon. This connection may be useful for
assessments of juvenile occupancy patterns.
How juvenile coho salmon come to occupy locations low
or high in the stream network is an important question. If the
pattern of juvenile occupancy reflects the spawning location of
their parents, then I can interpret the coupling of adult spawner
abundance and juvenile coho stream network occupancy to
suggest that more spawners use more stream kilometers for
spawning, hence, more stream kilometers are occupied by
juvenile coho salmon. If this is the case, then when adult runs
are low, spawning occurs successfully in lower stream reaches,
as reflected by lower juvenile network occupancy. It is also
interesting to note that higher numbers of stream kilometers
occupied by juveniles does not appear to be connected to
higher precipitation during the parental spawning run. If the
location of spawning adults helps define the upper distribution
of juvenile coho salmon, then this observation could mean that
adults are not necessarily moving high in the stream network
with high water during the spawning run, but are responding
to something else. Isaak and Thurow (2006) found that in
years of low abundance, Chinook salmon (O. tshawytscha)
spawned in specific core locations, expanding into other areas
as their abundance increased. I speculate that in years of low
abundance, adults did not return to their natal stream beds
further up the network in order to spawn. Perhaps access
to available spawning partners that could be found lower in
the stream network was a more important consideration for
spawning adults than migration further upstream.
Synthesis Among Scales
Salmonid population ecology identifies a wide variety of
behavioral mechanisms of both adults and juveniles that help
the fish cope with dynamic environmental conditions (Reeves
et al. 1995). Some needs for survival of coho salmon, such as
spawning gravels, summer pools and winter habitat have few
38
substitutes. However, individuals may modify their behavior
to take advantage of current stream and environmental
conditions. Isaak and Thurow (2006) used continuous
replicate surveys of stream networks over 9 years to describe
patterns in Chinook salmon redd distributions in Idaho. They
determined that core spawning areas were important in years
of low and high adult abundance, but maintaining additional
spawning areas in the network was important in years of
varying environmental conditions. Location in the stream
network becomes a critical contextual consideration for the
successful fulfillment of specific life history needs.
In this study, physical evidence of coho salmon behavioral
diversity and responsiveness to current population and
environmental conditions was found in the form of network
habitat topology and juvenile occupancy patterns across
multiple subbasins. Schlosser (1995) identified the importance
of large scale landscape conditions and the presence of refugia
as potentially important considerations of fish population
dynamics. These two considerations appeared relevant in
this study when all three spatial scales are considered. Site
scale differences in juvenile density among subbasins were
connected with the juxtaposition of necessary seasonal
habitats and network variables were found to be important in
modeling juvenile density. Patch sizes of juveniles varied with
subbasin and year, with potential nesting of juvenile clusters
in larger subbasins. This makes network connectivity and
subbasin size important contextual considerations for juvenile
subpopulations as individuals seek necessary seasonal habitats.
Further, subbasin occupancy patterns of juvenile coho salmon
support the idea that the location of habitats is not an obligate
consideration for spawning adults. Rather, adults may take
advantage of habitats with reference to the annual conditions
in which they find themselves.
An analysis of many subbasins with several spatial scales
highlights complexities in juvenile coho salmon density
and occupancy. The observed diversity of site scale instream
habitat topology and occupancy at the subbasin scale may be
important in considering the survival and adaptability of coho
salmon populations. Large scale variability has been shown
to influence the capacity of streams to produce brook trout
(Kocovsky and Carline 2006). The site scale observation that
several subbasins had either decreasing or increasing juvenile
density between 2001 and 2002 is made more interesting
when the observation at the subbasin scale of an almost
universal increase in stream kilometers occupied between
these two years is noted. If stream occupancy is associated
with spawning run size, then it may be possible that the site
scale observation of decreased density may indeed be related to
differential survival of juveniles. In this case, the identification
of juxtaposition variables as important in differentiating
between subbasins with increasing or decreasing coho density
Flitcroft
Chapter 3
shows the influence of habitat topology on survival even in a
region of similar topography.
Conclusions
There is more to an assessment of habitat than simple
calculations of habitat area and amount. The spatial
arrangement of habitats throughout the year and the
accessibility of habitats appear to be important in describing
distribution and abundance of juvenile coho salmon. In
particular, considering the spatial arrangement of habitats
and patches of juvenile coho salmon provided a spatial
context for the interpretation of multiple spatial scales that
represent different population and environmental processes.
Juvenile patterns of network occupancy appear to depend
on site-specific habitat, the proximity of necessary seasonal
habitats, adult dispersal through the spawning grounds and
spawning run size. All of these factors work together to create
the patterns of occupancy that are observed in the stream
network. Ecological analysis that incorporates multiple spatial
scales and processes may inform management actions that are
intended to preserve or restore populations of coho salmon.
Management actions that consider the interrelationships
between life history stages, habitats, and movement within
the stream network may be more effective at maintaining the
complex habitat configurations necessary for the persistence of
populations of coho salmon.
The varied response of juvenile coho salmon to the
complexity and diversity of stream habitats was mirrored
in patterns of occupancy and density across spatial scales.
The stream network became an important lens through
which to view juvenile density and provided connectivity of
interpretation across scales. Site-scale analysis that identified
the importance of juxtaposition variables informed the
interpretation of possible definitions of patch sizes at the
subbasin scale. Complex subbasin patterns of juvenile coho
salmon occupancy were shown to vary within and among
subbasins over time and suggested the importance of the size
of the adult spawning run.
Integrating the context of the stream network into analysis
at multiple spatial scales included an important dimension of
the environment in which coho salmon evolved, and juvenile
fish must survive. The riverscape in which coho salmon must
endure includes a diversity of habitats with varied productivity
and connectivity among years and environmental conditions.
Ultimately, the persistence of coho salmon depends on
access to multiple, diverse and connected habitat. Salmon
evolved the ability to embrace the diversity of habitats and
environmental conditions that naturally occur in the Pacific
Northwest. This adaptability is how they have coped with
the frequent disturbances and the inconsistent spawning and
39
rearing conditions that are present around the Pacific Rim.
Management strategies that fail to consider the
environmental context of the stream network ignore a critical
element of the riverscape for coho salmon. Depending on
small areas with historically high runs and juvenile survival is
not an adequate management strategy. Rather, management
strategies that consider the spatial distribution of habitats and
the variety of habitat topologies in different subbasins will
be better suited to the complex environmental conditions in
which coho salmon evolved and continue to persist.
Literature Cited
Bell, E, W.G. Duffy and T.D. Roelofs. 2001. Fidelity and survival of
juvenile coho salmon in response to a flood. Transactions of the
American Fisheries Society 130: 450-458.
Benda, L. and T. Dunne. 1997. Stochastic Forcing of Sediment
Supply to Channel Networks from Landsliding and Debris Flow.
Water Resources Research 33(12): 2849-2863.
Benda, L., L. Poff, D. Miller, T. Dunne, G. Reeves, G. Pess and M.
Pollock. 2004. The network dynamics hypothesis: How channel
networks structure riverine habitats. BioScience 54(5): 413-427.
Bisson, P.A., K. Sullivan and J.L. Nielsen. 1988. Channel hydraulics,
habitat use, and body form of juvenile coho salmon, steelhead,
and cutthroat trout in streams. Transactions of the American
Fisheries Society 117(3): 262-273.
Burnett, K.M. 2001. Relationships among juvenile anadromous
salmonids, their freshwater habitat, and landscape characteristics
over multiple years and spatial scales in the Elk River, Oregon.
Doctoral dissertation. Oregon State University, Corvallis, Oregon.
Burnett, K.M., G.H. Reeves, D.J. Miller, S. Clarke, K. VanceBorland and K. Christiansen. 2007. Distribution of salmon-habitat
potential relative to landscape characteristics and implications for
conservation. Ecological Applications 17(1): 66-80.
Chapman, D.W. 1962. Food and space as regulators of salmonid
populations in streams. The American Naturalist 100(913): 345357.
Clarke, S.E., K.M. Burnett and D.J. Miller. In review. Modeling
streams and hydrogeomorphic attributes in Oregon from digital
and field data. Journal of American Water Resources Research.
Collen, P. and R.J. Gibson. 2001. The general ecology of beavers
(Castor spp.), as related to their influence on stream ecosystems
and riparian habitats, and the subsequent effects on fish—A
review. Reviews in Fish Biology and Fisheries 10: 439-461.
Daly, C., G.H. Taylor, W.P. Gibson, T.W. Parzybok, G. L. Johnson
and P.A. Pasteris. 1999. High-quality spatial climate data sets
for the United States and beyond. Transactions of the American
Society of Agricultural Engineers 43(6): 1957-1962.
Day, D.M., W.A. Bertrand, M.J. Wiley and R. Sauer. 1992.
Influence of stream location in a drainage network on the index
of biotic integrity. Transactions of the American Fisheries Society
121: 635-643.
Flitcroft
Chapter 3
Dunham, J.B, B.E. Rieman and J.T. Peterson. 2002. Chapter 26:
Patch-based models to predict occurrence: lessons from salmonid
fishes in streams. Pages 327-334. in J.M. Scott, P. Heglund,
M. Morrison, editors. Predicting species occurrences issues of
accuracy and scale. Island Press, Washington, D.C.
Ettema, C.H. and D.A. Wardle. 2002. Spatial soil ecology. TRENDS
in Ecology and Evolution 17(4): 177-173
Fagan, W.F. 2002. Connectivity, fragmentation, and extinction risk
in dendritic metapopulations. Ecology 83(23): 3243-3249.
Fausch, K.D., C.E. Torgersen, C.V. Baxter and H.W. Li. 2002.
Landscapes to riverscapes: Bridging the gap between research and
conservation of stream fishes. Bioscience. 532(6) : 483-498.
40
habitats? Canadian Journal of Fisheries and Aquatic Sciences 60:
1527-1538.
Kahler, T.H., P. Roni and T.P. Quinn. 2001. Summer movement
and growth of juvenile anadromous salmonids in small western
Washington streams. Canadian Journal of Fisheries and Aquatic
Sciences 58: 1947-1956.
Kocik, J.F. and C.P. Ferreri 1998. Juvenile production variation in
salmonids: population dynamics, habitat, and the role of spatial
relationships. Canadian Journal of Fisheries and Aquatic Sciences
55(Suppl. 1): 191-200.
Flitcroft, R.L. 2007. Chapter 2: this volume. Oregon State
University, Corvallis Oregon.
Kocovsky, P.M, and R.F. Carline. 2006. Influence of landscapescale factors in limiting brook trout populations in Pennsylvania
streams. Transactions of the American Fisheries Society 135;
76-88.
Foster, S.C., C.H. Stein and K.K. Jones. 2001. A guide to
interpreting stream survey orts. Edited by P.A. Bowers. Oregon
Department of Fish and Wildlife, Information Reports 2001-06,
Portland, Oregon.
Labelle, M. 1992. Straying patterns of coho salmon (Oncorhynchus
kisutch) stocks from southeast Vancouver Island, British Columbia.
Canadian Journal of Fisheries and Aquatic Sciences 49: 18431855.
Frissell, C.A., W.J. Liss, C.E. Warren and M.D. Hurley. 1986. A
hierarchical framework for stream habitat classification: viewing
streams in a watershed context. Environmental Management
10(2): 199-214.
Lichatowich, J. 1999. Salmon without rivers a history of the Pacific
salmon crisis. Island Press. Coveto, California.
Ganio, L.M., C.E. Torgersen and R.E. Gresswell. 2005. A
geostatistical approach for describing spatial pattern in stream
networks. Frontiers in Ecology and the Environment 3(3): 138144.
Grant, J.W.A.., S.O. Steingrimsson, E.R. Keeley and R.A. Cunjak.
1998. Implications of territory size for the measurement and
prediction of salmonid abundance in streams. Canadian Journal
of Aquatic Science 55 (Supplement 1): 181-190.
Lonzarich, D.G., M.R. Lonzarich and J.L. Warren, Jr. 2000. Effects
of riffle length on the short-term movement of fishes among
stream pools. Canadian Journal of Fisheries and Aquatic Sciences
57: 1508-1514.
Matthews, W.J. and E. Marsh-Matthews. 2003. Effects of drought
on fish across axes of space, time and ecological complexity.
Freshwater Biology 48: 1232-1253.
McCune, B. 2004. Nonparametric multiplicative regression for
habitat modeling. Available: http://www.pcord.com/NPMRinto.
pdf. (January 2007).
Groot, C. and L. Margolis editors. 1991. Pacific Salmon Life
Histories. University of British Columbia Press, Vancouver, B.C.,
Canada.
McCune, B. and M.J. Mefford. 1999. PC-ORD. Multivariate
analysis of ecological data. Version 4.33 MjM Software, Gleneden
Beach, Oregon.
Isaak, D.J and R.F. Thurow. 2006. Network-scale spatial and
temporal variation in Chinook salmon (Oncorhynchus tshawytscha)
redd distributions: Patterns inferred from spatially continuous
replicate surveys. Canadian Journal of Fisheries and Aquatic
Sciences 63: 285-296.
McCune, B. and M.J. Mefford. 2004. Hyperniche. Nonparametric
multiplicative habitat modeling. Version 1.05 MjM Software,
Gleneden Beach, Oregon.
Jacobs, S., J. Firman, G. Susac, D. Steward and J. Weybright. 2002.
Status of Oregon coastal stocks of anadromous salmonids, 20002001 and 2001-2002. Oregon Department of Fish and Wildlife,
Report OPSW-ODFW-2002-3, Portland, Oregon.
Jacobs, S.E. and R.E. Nickelson. 1989. Use of stratified random
sampling to estimate the abundance of Oregon coastal coho
salmon. Oregon Department of Fish and Wildlife, Report
F-145-R-09, Portland, Oregon.
Johnson, S.L. 1999. Fish habitat restoration in tenmile Creek lifecycle monitoring project summary report. Oregon Department of
Fish and Wildlife, Salem, Oregon.
Jones, M.L., J.K. Netto, J.D. Stockwell, and J.B. Mion. 2003.
Does the value of newly accessible spawning habitat for walleye
(Stizostedion vitreum) depend on its location relative to nursery
Megalhaes, F.M., D.C. Batalha and M.J. Collares-Pereira. 2002.
Gradients in stream fish assemblages across a Mediterranean
landscape: Contributions of environmental factors and spatial
structure. Freshwater Biology 47: 1015-1031.
Miller, D.J. 2003. Programs for DEM analysis. In landscape
dynamics and forest management. General Technical Report
RMRSGTR-101CD. USDA Forest Service, Rocky Mountain
Research Station, Fort Collins, Colorado.
Miller, D.J., C. Luce and L. Benda. 2003. Time, space, and
episodicity of physical disturbance in streams. Forest Ecology and
Management 178: 121-140.
Montgomery, D.R., E.M. Beamer, G.R. Pess and T.P. Quinn. 1999.
Channel type and salmonid spawning distribution and abundance.
Canadian Journal of Fisheries and Aquatic Sciences. 56: 3770387.
Flitcroft
Chapter 3
Moore, K.M.S., K.K. Jones and J.M. Dambacher. 1997. Methods
for stream habitat surveys. Oregon Department of Fish and
Wildlife, Information Report 97-4, Portland, Oregon.
Naiman, R.J., T.J. Beechie, L.E. Benda, D.R. Berg, P.A. Bisson,
L.H. MacDonald, M.D. O’Connor, P.L. Olson and E.A. Steel.
1992. Fundamental elements of ecologically healthy watersheds
in the Pacific Northwest coastal ecoregion. Pages 127-187 in R.J.
Naiman, editor. Watershed Management: Balancing Sustainability
and Environmental Change. Springer-Verlag, New York.
Nakamura, F., F.J. Swanson and S.M. Wondzell. 2000. Disturbance
regimes of stream and riparian systems – A disturbance-cascade
perspective. Hydrological Processes 14: 2849-2860.
Nickelson, T.E., M.F. Solazzi, S.L. Johnson and J.D. Rodgers.
1992a. An approach to determining stream carrying capacity and
limiting habitat for coho salmon Oncorhynchus kisutch. Pages 251260 in L. Berg and P.W. Delaney, editors. Proceedings of the coho
workshop. Naniamo, B.C., Canada.
Nickelson, T.E., J.D. Rodgers, S.L. Johnson and M.F. Solazzi.
1992b. Seasonal changes in habitat use by juvenile coho salmon
Oncorhynchus kisutch in Oregon coastal streams. Canadian Journal
of Fisheries and Aquatic Sciences 49(4): 783-789.
Nielsen, J. 1992. Microhabitat-specific foraging behavior, diet, and
growth of juvenile coho salmon. Transactions of the American
Fisheries Society 121: 617-634.
Oregon Department of Fish and Wildlife. 2005. Oregon list of
threatened and endangered fish and wildlife species. Available:
http://www.dfw.state.or.us/
threatened_endangered/t_e.html
(July 2007).
Peters, D.P.C., J.R. Gosz, W.T. Pockman, E.E. Small, R.R.
Parmenter, S.L. Collins and E. Muldavin. 2006. Integrating
patch and boundary dynamics to understand and predict biotic
transitions at multiple scales. Landscape Ecology 21: 19-33.
Pollock, M.M., G.R. Pess and T.J. Beechie. 2004. The importance of
beaver ponds to coho salmon production in Stillaquamish River
Basin, Washington, USA. North American Journal of Fisheries
Management 24: 749-760.
Quinn, T.P. and N.P. Peterson. 1996. The influence of habitat
complexity and fish size on over-winter survival and growth of
individually marked juvenile coho salmon (Oncorhynchus kisutch)
in Big Beef Creek, Washington. Canadian Journal of Fisheries and
Aquatic Science 53: 1555-1564.
Redmond, K. and G. Taylor. 1997. Chapter 2 – Climate of the
coastal temperate rain forest. Pages 25-42 in Schoonmaker, P.K.,
B. von Hagen and E.C. Wolf editors. The Rain forests of Home:
Profile of a North American Bioregion. Island Press, Washington,
D.C.
Reeves, G.H., L.E. Benda, K.M. Burnett, P.A. Bisson and J.R. Sedell.
1995. A disturbance-based ecosystem approach to maintaining
and restoring freshwater habitats of evolutionarily significant units
of anadromous salmonids in the Pacific Northwest. American
Fisheries Society Symposium 17: 334-349.
Remington, K. 1999. ArcScript: shortest network paths v1.1.
Available:
http://arcscripts.esri.com/details.asp?dbid=10895.
41
(January 2007).
Rosenfeld, J., M. Porter, E. Parkinson. 2000. Habitat factors
affecting the abundance and distribution of juvenile cutthroat
trout (Oncorhynchus clarki) and coho salmon (Oncorhynchus
kisutch). Canadian Journal of Fisheries and Aquatic Sciences 57:
766-774.
Schlosser, I.J. 1985. Flow regime, juvenile abundance, and the
assemblage structure of stream fishes. Ecology 66(5): 1484-1490.
Schlosser, I.J. 1991. Stream fish ecology: A landscape perspective.
BioScience 41(10): 704-712.
Schlosser, I.J. 1995. Critical landscape attributes that influence fish
population dynamics in headwater streams. Hydrobiologia 303:
71-81.
Sleeper, J. 1993. Seasonal changes in distribution and abundance
of salmonids and habitat availability in a coastal Oregon basin.
Masters thesis. Oregon State University, Corvallis, Oregon.
Strahler, A.N. 1952. Dynamic basis of geomorphology. Geological
Society of America Bulletin 63: 923-938.
Swanson, F.J., S.L. Johnson, S.V. Gregory and S.A. Acker. 1998.
Flood Disturbance in a Forested Mountain Landscape Interactions
of land use and floods. BioScience 48(9): 681-689.
Swift, L.W., G.B. Cunningham, and J.E. Douglass. 1988.
Climatology and hydrology. In W.T. Swank and D.A. Crossley,
Jr., eds. Ecological Studies, Volume 66: Forest Hydrology and
Ecology at Coweeta, pp. 35-55, Springer-Verlag, New York.
Talley, T.S. 2007. Which spatial heterogeneity framework?
Consequences for conclusions about patchy population
distributions. Ecology 88(6): 1476-1489.
Torgersen, C.E., R.E. Gresswell and D.S. Bateman. 2004. Pattern
detection in stream networks: Quantifying spatial variability in fish
distribution. In T. Nishida, P. J. Kailola and C.E. Hollingworth,
eds. GIS/Spatial Analyses in Fishery and Aquatic Sciences, pp.
405-420, Fishery and Aquatic GIS Research Group, Saitama,
Japan.
Trombulak, S.C. and C.A. Frissell. 2000. Review of ecological effects
of roads on terrestrial and aquatic communities. Conservation
Biology 14(1): 18-30.
United States Geological Survey. 2005. NWISWeb: New site for the
nation’s water data. Fact Sheet 128-02.
Wiens, J.A. 1989. Spatial scaling in ecology. Functional Ecology 3:
385-397.
Wimberly, M.C., T.A. Spies, C.J. Long and C. Whitlock. 2000.
Simulating historical variability in the amount of old forests in
the Oregon Coast Range. Conservation Biology 14(1): 167-180.
Wright, K.K. and J.L. Li. 2002. From continua to patches:
Examining stream community structure over large environmental
gradients. Canadian Journal of Fisheries and Aquatic Sciences 59:
1404-1417.
Flitcroft
Chapter 3
42
Appendix 3.1: ODFW Aquatic Inventory and Analysis Project: Habitat Benchmarks.
Pools
POOL AREA (% Total Stream Area)
POOL FREQUENCY (Channel Widths Between Pools)
RESIDUAL POOL DEPTH (m)
SMALL STREAMS (<7 m width) MEDIUM STREAMS (≥ 7m and < 15 m width)
LOW GRADIENT (slope <3%) HIGH GRADIENT (slope >3%) LARGE STREAMS (≥15m width) COMPLEX POOLS (Pools w/ LWD pieces ≥3) / km RIFFLES
WIDTH / DEPTH RATIO (Active Channel Based)
EAST SIDE WEST SIDE GRAVEL (% AREA) SILT-SAND-ORGANICS (% AREA)
VOLCANIC PARENT MATERIAL SEDIMENTARY PARENT MATERIAL CHANNEL GRADIENT <1.5% SHADE (Reach Average, Percent)
STREAM WIDTH <12 meters
WEST SIDE NORTHEAST CENTRAL - SOUTHEAST STREAM WIDTH >12 meters
WEST SIDE NORTHEAST CENTRAL - SOUTHEAST Undesirable
<10
>20
Desirable
>35
5-8
<0.2
>0.5
<0.3
<0.5
<0.8
<1.0
>0.6
>1.0
>1.5
>2.5
>30
>30
<15
<10
<15
≥35
>15
>20
>25
<8
<10
<12
<60
<50
<40
>70
>60
>50
<50
<40
<30
>60
>50
>40
<10
<20
<1
>20
>30
>3
<150
<75
>300
>200
LARGE WOODY DEBRIS* (15 cm x 3 m minimum piece size)
PIECES / 100 m STREAM LENGTH VOLUME / 100 m STREAM LENGTH “KEY” PIECES (>60 cm dia. & ≥10 m long)/100 m RIPARIAN CONIFERS (30 m FROM BOTH SIDES CHANNEL)
NUMBER >20in dbh/ 1000ft STREAM LENGTH NUMBER >35in dbh/ 1000ft STREAM LENGTH
*Values for streams in forested basins
Citation:
Foster, S.C., C.H. Stein and K.K. Jones. 2001. A guide to interpreting stream survey reports. Edited by P.A. Bowers. Oregon Department
of Fish and Wildlife, Information Reports 2001-06, Portland.
Flitcroft
Chapter 3
43
Appendix 3.2. Variograms for small basins in Oregon’s Mid-Coast Region in each year available between 1998 and 2002 (Figure Series
3.2.1-3.2.5.).
Figure 3.2.1. Canal Creek Variograms from 1998, 2001, and 2002.
Figure 3.2.2. Cedar Creek Variograms from 1999, 2001, and 2002.
Flitcroft
Chapter 3
Appendix 3.2. (continued).
Figure 3.2.3. Sams Creek Variograms from 1999, 2001, and 2002.
Figure 3.2.4. Fall Creek Variograms from 1998, 2001, and 2002.
44
Flitcroft
Chapter 3
Appendix 3.2. (continued).
Figure 3.2.5. Sunshine Creek Variograms from 1998, 1999, 2001, and 2002.
45
Flitcroft
Chapter 3
46
Appendix 3.3. Maps (1b-9d) of juvenile coho salmon density in eleven basins of Oregon’s Mid-Coast in selected years from 1998 through
2002.
1b. Drift Creek, 1999
#
##
#
S
#
#
S
#
##
#
S
#
#
S
##
#
#
#
S
#
#
#
S
#
#
#
#
#
#
# #
S
#
#
#
#
#
S
#
#
S
#
#
##
S
#
#
#
#
#
#
S
S
S
#
#
#
#
S#
#
S
#
#
#
S
S
##
#
#
S
#
S
#
#
#
#
## #S##
##
#
S
#
## #
#
S#
S
#
#
S#
S
#
Juvenile Coho
Salmon Density
##
#
##
#
#
#
#
#
SS
S
## #
#
#
#
#
S
S
#
S
#
S
#
S
S#
#
#
##
S#
#
#
#
S
S
S
#
S
##
S
#
S
#
#
##
S
#
#
S
#
S
#
#
S
#
#
S
#
S
#
#
#
S
S
S
S
##
##
S
S
S ##
#S
#
S
###
# #
#
##
S
#
S
#
#
#
###
# #
0
0.01-0.50
#
S
#
S
#
#
S
#
S
#
S
#
#
#
#
#
#
S
S
#
S
#
#
#
S
S##
#S
0.51-1.0
1.01-2.0
2.01-10.0
1c. Drift Creek, 2001
#
#
#
#
##
#
#
#
#
# #
#
#
#
#
#
#
#
#
#
##
##
#
#
#
###
S#
#
#
# #
#
#
S
#
##
#
#
S
#
#
#
#
S#
#
#
#
#
#
#
#
#
#
#
#
S#
S
#
#
S
S
#
#
##
#
S
#
#
S
#
#
#
#
#
#
#
#
#
#
#
#
#
S
##
#
#
#
#
S
#
S
#
#
S
#
#
S
#
#
#
#
#
#
#
S#
S
#
#
#
#
#
S#
##
#
#
#
##
#
#
#
#
S
S
S#
#
#
S
#
#
#
S
S
#
S
#
#
#
S
#
#
S
S
#
#
#
#
##
#
#
S
# ##S#S#S#S#S
#
#
#
#
#
#
#
#
S
#
#
#
#
#
S
#
#
#
#
S
#
S
##
#
#
S
#
#
S
#
##
#
#
#
S
#
#
S
#
S
#
##
#
S
#
SS
#
##
#S
#
S#
#
S
#
#
S
#
S
##
#
S
#
S #
#
S##
##
S
S
#
SS#
S
S#
S#
##
## ##
##
#
#
###
S
S
S##
#
S
##
##
#
S
#
S
S
#
#
1d. Drift Creek, 2002
##
# #
#
#
#
#
#
S
#
#
#
S #
#
##
#
#
#
#
#
#
#
#
#
#
#
#
#
##
S
#
#
##
#
#
S
S
###
#
#
S
##
#
#
#
S ##
#
S
S
#
#
#
S
#
#
S
S
#
#
S
S#
#
#
#
S#
#
#
S
#
#
S
#
#
#
#
#
#
S
#
#
#
S#
S
#
#
S
#
#
S
#
#
S
#
S
S
#
#
##
S
#
#
#
#
SS
#
#
#
#
#
#
#
#
#
#
#
#
#
S
S
#
S#
##
#
S #
#
#
#
#
S
S
#
S
#
S
S
#
S
#
#
S
#
S
#
#
#
#
#
S
S
#
S#
S
#
##
#
S
##
S
#
#
#
S
#
S#
#
#
#
S
#
S
#
#
#
#
#
S
##
S
S
#
#
#
#
S
S
S
S
#
#
S
#
#
#
S
#
#
#
#
#
#
#
#
#
S
S
S
#
#
#
#
#
S
S
#
#
#
S
##
# ###S#S#S##S
#
#
#
#
#
##
S#
#
#
#
#
S
#
#
#
S
#
S
#
#
#
#
S
S
S
S
#
#
#
#
S
#
#
S
S
#
#
#
#
#
S
#
S
#
#
#
S
#
S#
S
S
#
#
S
S#
#
S
#
S
#
#
S #
SS#
#
S #
#
S
# ###
SS#
#
S#
S
S##
#
S
S
S
S
S#
#
#
S
##
S
#
S
#
S
#
#
#
S
S
#
#
S
#
#S
#
#
#
S#
S
#
S
#
S
#
#
S
S
#
#
S
##
#
S
#
S
#
#
S
S
#
S
#
#
#
S
#
Flitcroft
Chapter 3
47
Appendix 3.3. (continued)
2a. Fall Creek, 1998
#
#
S
#
#
#
S
#
S
###
#
#
S
#
#
S#
#
S
S#
S
###
#
S#
#
S#
S
#
S
###
#
S
#
S#
#
S
S#
#
S
S
S##
#
##
#
#
S#
SS
S
#
S
#
S
#
S
#
S
#
S
#
S#
#
S
S
#
S
#
S
#
S
#
S#
S
#
S
#
S
#
S
#
S
#
S
SS
#
#
S#
#
S
#
##
S
#
S
#
S#
#
S
#
S#
#
#
S
#
#
S
#
S#
S
S##
S#
#
#
#
S
#
#
S
#
#
S# #
S
#
S##
###
S
S##
#
Juvenile Coho
Salmon Density
0
0.01-0.50
0.51-1.0
1.01-2.0
2.01-10.0
#
# ### # #
#
#
#
2c. Fall Creek, 2001
2d. Fall Creek, 2002
#
#
#
S
##
#
S##
S#
S#
#
S#
#
##
#
S
#
S#####
#
##
#
S
S
#
S#
#
##
S
#
#
#
S
S
#
S#
S
#
#
S#
S
#
#
S
#
S
#
S
#
S
#
#
#
S
#
S
#
S
#
S##
#
S
#
#
S
#
S
#
#
S
#
S#
#
#
#
S#
S
#
S#
#
S
##
#
S
#
#
##
#
#
#
##
S#
#
S#
S#S ##
#
#
S#
S
#
S
#
S
#
#
S
##
#
#
###
##
##
S#
S#
S
#
#
#
#
S
#
S#
S#
S
S
#
S
S#
#
S#
#
S#
S
#
#
###
#
#
S
#
S
#
S#
#
S#
S
#
S
#
S
#
S
S
#
S
#
S#
#
S
####
S
#
S#
S#
S##
##
S
S#
#
S
#
S###
S
#
S#
##
#
#
##
#
#
#
###S#S#S#S#
####
####
# #
# #
S
#
S#
#
S#
#
#
#
#
Flitcroft
Chapter 3
48
Appendix 3.3. (continued)
3a. Five Rivers, 1998
3b. Five Rivers, 1999
#
#
S###
#
S
##
S
#
#
#
S
###
S
S
#
#
S
Juvenile Coho
Salmon Density
0
#
0.01-0.50
#
#
#
#
S
#
S#
#
#
S#
S
S ##
#
S
S##
#
###
S#
S#
#
#
#
#
S
S# #
S#
#
S#
##
S##
#
S#
S
S#
S#
S
S####
#
#
#
S
S
#
#
#
#
#
#
S
S
S
S
S
S
#
#
##S#S#S #S##S#S##S##S##S######
#
#
#
S
#
#
## #####
S
#
#
S #
#
#
S
S
#
#
#
#
#
S
S
S
#
#
S
S
#
#
#
SS
#
#
#
#
#
# #
S
#
#
#
#
#
#
#
S
S
S
S
#
#
#
#
S
#
#
#
#
S
S
S
S
S
#
#
#
S
S
#
#
#
#
#
S
S
S
S
#
#
S
S
S
#
#
#
#
S
S
#
#
#
S
#
#
#
#
#
##
#
#
S
S
#
#
S
S
##
#
S
#
#
S
#
#
#
#
# ###
S
#S##
#
#
S
S
#
#
S
#
#
#
#
S
##
##
S
#
#
S
#
#
S
S
#
#
#
#
#
#
S
#
#
S
#
#
#
S
S
#
S# ##
#
#
#
S
S
#
#
S
#
#
S
S
#
#
#
#
#
S
#
#
S
S#
#
#
#
#
S#
S
S
S#
#
S
S#
S
S#
#
#
S
S
#
#
S#
S#
#
#
S
S#
S#
#
S
#
#
S
#
#
#
S
S
S
#
#
#
S
#
#
S
S
#
#
#
S
S
#
S
#
#
#
S
S
#
#
S
#
#
S
#
#
S
S
#
#
#
#
S
#
#
#
#
#
S
S
S#
##
#
#
#
#
S#
S
S
#
#
S
#
#
#
#
S
#
#
S
S
#
#
#
#
S#
#
#
S
S
S
#
#
S
#
#
S
#
#
S
#
#
#
#
#
S
S
#
#
S
S
S#
#
S
##
#
S
#
#
S
#
#
#
#
S
#
#
#
S
S
#
S
#
#
S
#
#####
#
#
S
S
S
#
S
#
#
#
S
S#
#
#
#
S
S
#
#
#
#
S
S
S
#
S#
S
#
S
#
S#
#
#
S#
S
#
S#
#
S#
S#
#
#
#
#
S
S
#
S
S
#
S
#
S
#
#
#
S
#
S
#
#
S
#
S
#
S
#
S#
# #
S
#
#
#
S
#
#
#
#
S
S
#
#
#
#
#
#
S
S
S
S
S
S
S
#
#
S
##
#
#
#
S
#
#
S
#
#
S
#
#
S#
#
#
#
S
S
#
#
#
##
S
S
#
#
#
#
S
#
#
#
#
#
S
#
S
S
S
#
#
S
#
S
#
#
#
#
S
S
S
#
S
#
#
#
S
#
#
S
#
#
#
#
S
S
S
S
#
#
#
S
#
#
#
#
S
S
S
#
#
#
#
#
#
S
S
#
#
S
#
#
#
#
#
#
#
S
#
#
S
#
#
S
##
#
S
#
S
#
##
S
#
#
##
#
#
#
#
S
#
S
#
#
#
#
S
#
S
#
#
#
#
#
#
S
#
#
##
#
S
#
#
#
#
S
#
#
#
#
S
0.51-1.0
1.01-2.0
2.01-10.0
##
#
#
S
#
#
S
#
#
S
#
S
#
#
S
#
S
#
S#
#
S
#
#
S
#
S
#
S
#
#
#
#
#
S
S
#
#
#
S
# #
S
S
S
#
S
#
#
S
#
#
S
S
#
S
#
SS #
S
S
#
#
###
S
#
#
#
S
#
S
#
S
#
S
#
#
#
#
#
S
#
#
#
#
#
#
#
#
S
#
#
S
S
#
S
#
###
#
S
#
#
#
##
#
3c. Five Rivers, 2001
3d. Five Rivers, 2002
###
##
#
S#
##
S
#
S
#
S ###
#
#
S
#
S
#
#
S
#
#
##
S
#
###S#S##
#
#
#
#
#
S
#
S
#
SS
#
#
S
#
#
#
S #
S
#
S
#
S
##
S
S##
#
SS#
S
#
S#
S#
S#
#
#
S#
S#
###
S#
##
##
S
S
#
#
S
SS#
S
#
#
S
S
#
S#
#
#
#
S
S
S
#
#
S
###
S
#
S
#
S
#
#
S
## S
#
S
#
S
#
#
S
#
S#
#
S
S
S#
#
#
S
#
# #
S##
S
S #
#
#
S
#
#
#
S
S
S
S
#
#
#
S
#
S
#
#
#
#
SS
S
#
#
S #
S
S
S
#
#
#
#
S
#
S
#
S
#
S
#
#
SS
S
#
#
S#
#
#
#
S S
S#
S
#
S#
S
#
#
S
#
#
S
S
#
#
#
S
S
#
S#
S#
#
S
S #
S#
#
S#
#
S
#
#
S#
S#
#
S
SS
##
S
#
#
#
#
S#
#
#
S#
#
#
S
S
##
#
#
S
#
S
#
S#
S
#
#
#
S
S#
SS
#
S
#
##
S
#
S
S
##
S#
#
S
#
#
#
S
S#
S
##
S
#
##
#
S
#
#
S#
S
#
S
#
#
#
#
S
#
#
S
S
#
#
S
#
S
#
#
#
S
#
#
#
#
#
#
#
###
#
S
S
#
#
#
#
S
#
#
S
#
#
S
S
#
#
#
#
#
#
S
#
S
S
#
#
#
S#
S
#
#
S
S
#
#
#
#
S
#
#
#
S
# #S
#
S#
#
#
S
S
#
#
#
S
S
#
#
#
#
S
#
S
S
S
#
#
S
#
S#
#
#
#
S
#
S
## ###S#
#
S
##
#
S
#
#
S#
S
####
##S
S
S#
#
#
#
#
S
S#
#
#
#
S
S
#
#
S
# #
#
#
#
S
##
#
##
#
#
#
#
S
#
S
#
#
#
S
#
#
S
##
#
#
#
#
S
#
#
#S
#
#
#
##
#
#
#
#
S
#
S
S
#
S#
##
#
#
#
#
#
#
S
S
S
S
S
#
#
#
#
#
S
#
S
#
S
#
#
#
#
#
#
S
#
#
#
##
#
#
#
#
#
##
#
#
##
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
# # ##
#
#
#
####
#
###
#
#
##
#
#
#
##
##
#
#
##
#
#
#
#
#
#
####
#
#
#
#
##
#
#
##
#
###
#
#
#
#
#
#
#
#
#
#
#
##
#
#
##
#
#
#
#
#
#
S
#
S
#
#
S
#
SS#
S#
##
S
#
S#
#
S
#
S
#
S
S#
S
# ##
##
#
####
S
#
S#
S##
##
#
#
S
#
#
#
#
#
#
#
#
##
S
S
S
#
S
S
S
S
S
##
#
S
#
#S
#
#
#
S
S
#
#
S
S
S#
#
S
S
S
S
#
#
S
#
SS
#
S #
#
#
S#
#
S
S#
#
S#
#
#
###
S#
S#
S
#
S
S
S
#
#
S
S
#
#
#
S
#
#
S
S
S
S
#
#
S
#
#
S#
S
SS
#
S
#
S
#
#
#
S
#
S
#
S
#
#
S
## #
#
#
S #
S
S
#
S#
#
#
S
#
#
S
S#
#
S
S
#
#
S
#
#######
#
S
S
S #
#
#
S
S#
#
#
S
S
#
#
#
S
S
S #
#
S
#
#
#
#
#
S
S
S
#
#
#
S
S
#
S
#
#
#
S
#
S#
#
#
S
S
S
#
#
#
#
S
S
S
S#
S#
## #
#
S
#
S
#
S #
S####
#
S
S
#
#
S
#
#
S
#
S S #
S
#
#
S
#
#
#
#
S #
#
#
S
#
#
S
#
S
#
##
S
S
#
S#
#
#
S#
#
#
##
S
S
#
S
#
S
S
S
#
#
#
S#
#
#
S
#
SS
S
S
S
#
#
S
#
#
#
S
#
#
##
S
S
#
S
S#
#
#
S#
# #
#
S
#
#
#
S
#
#
S
S
S
#
#
S
#
S
S
#
S
#
#
#
#
#
S
S
#
S
#
#
#
#
#
#
#
#
S
S
S
#
#
#
S
S
#S#S#SS
SS
##
##
#
#S
S
#
#
#
#
S#
#
#
#
S#
S#
#
#
#
#
S #
#
#
#
S
#
#
#
#
#
S
#
S#
S
#
S
#
#
#
S #
#
S
#
#
#
S
#
# #
#
S
#
S
#
S
#
##
#
S#
#
#
#
##
S#
#
##
S
#
#
S
#
S#
#
S#
#
S
#
#
S
#
#
S
S
S
#
S
#
#
#
#
S
#
S
#
#
S
S
#
#
#
##
S#
S
S #
#
S
S
#
#
##
#
S
S
#
#
#
#
S#
#
S
##
#
##
S
#
#
#
#
S
#
#
S
S
#
#
S
#
#
S
#
#S##
#
#
S
#
#
#
##
#
#
#
#
#
#
#
S###
S
#
#
#
#
#
#
#
#
#
#
S
S
#
#
#
#
S
#
#
#
##
#
#
##
#
#
#
##
##
#
##
##
###
###
#
#
#
#
#
#
#
#
#
##
#
### #
#
#
#
###
#
#
#
#
#
#
#
#
#
#
#
#
#
# #
#
#
#
#
###
#
#
##
#
#
#
#
#
#
#
#
#
#
##
## #
##
#
##
#
##
##
#
##
#
#
#
#
#
## #
#
#
#
#
#
#
#
#
#
#
#
##
#
#
#
#
#
#
#
# #
#
#
#
####
#
#
#
#
#
#
##
## #
#
#
#
#
##
#
##
#
#
#
#
#
#
#
#
#
#
#
#
#
##
#
##
##
#
#
#
## #
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
S
S
#
#
#
#
#
#
#
#####
Flitcroft
Chapter 3
49
Appendix 3.3. (continued)
4b. Lobster Creek, 1999
4a. Lobster Creek, 1998
##
#
## # ##
##
S
##
S
#
S
# #
##
#
#
S
#
S
#
#
#
S
# #
S
##
#
#
#
S
# #### #
S
#
S
S
#
#
S
S
#
#
#
#
S
S
S
#
#
#
#
S
S
S
#
# #
#
#S#
#
S
#
#
S
#
S
#
S
#
S
#
S # ##
#
#
#
S
#
S
S
#
##
# ##
S
S#
#
#
S
SS
##
S
#
##
##
#
#
S
#
S
S##
S##
S#
S
S#####
#
S
#
#
S
#
##
S
S
#
#
#
#
#
#
S#
S#
S
S
#
#
#
#
S#
S
S
S
S
#
#
S
#
#
S
#
#
#
S#
S
S
#
#
#
S
S
S
#
S
S
#
S
S
S
#
#
#
S
#
S
#
S
#
S
#
#
S
S
#
#
#
#
#
S##
S
#
#
# S
#
#
S
#
#
#
S
#
#
#
#
#
S
S
#
#
S#
# ## # #
#
S
#
S
S
#
##
S## #
##
#
#
###
S
#
S
S
#
S
S#
#
S
# #
#
S
#
S
#
S
#
S
#
#
#
#
#
S
#
S
S
#
S
#
S
#
S
#
S
#
#
S
S
#S
S
#
#
S
#
#
S
##
#
#
#
S
#
#
#
S
#
S
S
#
#
S
#
#
S
#
S
#
S
#
# ##
S
#
#
S
S
S
#
#
#
S
S
#
#
S
#
#
#
S
S
#
#
S
#
#
S
#
#
#
S
S
S
#
S
#
#
S
#
S
#
S##
S
#
#
S
#
##
#
#
#
#
##
##
#
#
#
#
#
#
#
#
#
###
#
#
#
#
#
#
#
##
# # #
#
#
#
#
##
#
# #
#
#
#
#
#
#
##
#
##
#
#
##
##
#
######
#
#
### # # #
#
S##
#
#
S
##
#
#
S
#
S
# #
#
S
## #
#
#
#
# #
#
#
S
#
#
S
S# #
## # #
S##
#
##
#
#
S
#
#
#
S
S
#
S
#
S
#
S#
S
#
#
#
S
#
#
S
#
S#
#
S
#
S
S
#
#
S####
#
#
S
S###
#
S###
S#####
##
#
S
#
#
#
#
#
#
#
#
S
#
S#
S
S
S
S
S
#
#
S
S
S#
S
S
#
#
#
S
S
#
#
#
S
#
S
S
S
#
S
S
#
#
S
S
#
#
S
S
###
#
S
#
S
#
#
S
S
S
#
S##
#
S
#
S
#
#
S
#
S
#
#
S
#
#
#
S
#
S
#
##
#
S##
###
#
S
#
S
#
S
#
#
#
S
S
Juvenile Coho #
#
# #
#
#
S
S
#
#
S
S
#
S#
S
S
#
#
S#
#
S
#
#
#
#
#
#
#
##
#
#
#
## # ##
##
#
#
#
####
#
#
#
#
# # #
#
###
#
#
S
S
#
S#
S
##
#
Salmon Density
#
S
#
S
S
#
S
##
#
S
##
#
S
#
#
S
####
#
S
#
#
###
0
#
##
##
#
S
#
#
#
#
S
#
S
#
#
#
#
#
S
#
S#
S#
##
S
#
S#
S
#
#
#
S
#
S
S
S
S
#
#
S
#
S
#
S#
##
#
#
#
0.01-0.50
0.51-1.0
1.01-2.0
2.01-10.0
4c. Lobster Creek, 2001
S #
#
S
##
##
#
S
S
#
S##
#
SS #
##
#
S##
S
#
S
#
#
S#
S#
#
#
## #
S
SS#
SS#
S#
S# #
#
S
#
#
#
#
S#
S
#
S
#
#
# ###
#
S
S
# #
## #
#
#
S
##
#
#
#
#
S
#
##
S
##
S
#
#S
S
S
#S
#
#
#
S#
S#
S
S
#
S
S#
SS
S#
#
#
#
S
#
S
#
#
#
#
#
#
#
#
#
#
#
#
S
S
#
S
S
S
#
#
S
S
S
S
S
S
S
S
#
S
#
#
#
##
S
S
#
#
#
S
S
S
####
##
###
##
#
#
##
##
# #
#
#
S
#
#
S
#
#
S
#
S
#
#
##
#
S #
#
S
S#
S#
SS#
S#
#
#
S
S#
#
#
S
#
S
#
S
####
S#
#
#
#
S
S
####
#
# ### #
#
##
#
#
#
#
#
#
#
# ##
#
##
#
#
#
#
#
#
##
#
#
#
#
#
##
## #
#
#
#
#
#
#
##
#
#
#
#
##
#
#
#
#
#
#
#
#
#
#
#
#
#
##
#
S
#
#
#
#
#
S
##
#
#
#
#
#
#
#
#
#
#
#
#
#
#
##S##
#
#
#
#
#
#
#
#
#
S
#
#
4d. Lobster Creek, 2002
#
#
S#
# #
#
#
S
#
S
#
#
#
S SS
#
S#
S
#
#
S
#
S #
#
S
#
S
S#
#
##
##
S#
S
S##
#
#
S
#
#
#
S
S
#
#
S
#
#
S
#
#
S
S#S#
#
#
S# #
#
#
S
S
#
#
S#
#
S
#
#
#
S#
#
S
#
#
#
#
#
S
S
S
#
#
# S#
#
S#
S
S
S#
#
S
S#
S#
S
#
#
#
S
S
S
#
#
#
#
#
#
#
#
S
S
#
#
#
S
S
S
#
S
#
#
#
#
#
#
S
S
S
S
S
S
S
S
S
S
S
#
#
S
#
S
#
S
S
#
S
#
#
#
#
S
S
S
#
#
#
###
S #
#
#
#
#
#
###
S
S
S
S
#
#
##
##
#
S
#
#
#
#
S
##
##
#
#
#
#
S
#
###
#
##
#
#
S
S
##
S
#
S
#
#
#
#
S
S
S#
#
##
#
S
#
#
S
S
#
#
#
S
# #
#
S
#
S
#
#
#
#
S
#
#
#
#
#
S
S
#
#
#
S
#
S
#
#
S
# ##
#
#
S
S
#
S
#
S
#
##
#
S
#
#
#
S
# #
#
####
S
S
#
#
S
##
#
#
#
S
#
#
##
S#
#
#
S
##
#
#
S
#
#
#
##
#
#
#
# #
#
#
#
#
#
#
#
#
#
#
#
#
#
#
##
#
#
#
#
#
##
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
# #
#
#
#
##
#
##
#
#
#
#
#
# ##
#
##
#
#
#
#
#
##
#
##
#
#
#
#
#
#
#
#
Flitcroft
Chapter 3
50
Appendix 3.3. (continued)
5a. North Fork Alsea River, 1998
###
#
##
#
#
S#
S
#
S
#
S
S#
S
S#
#
S
#
#
S
S
S#
#
##
####
#
#
S
###
#
S
#
#
S
#
#
#
S
#
#
#
S#
###
#
S
#
#
##
#
#
#
#
S
#
#
#
#
S
#
#
S
##
#
S#
#
#
S
#
##### ### ##
S
##
# ##
##
#
#
S
#
S
#
S
# #
S
#
S
##
S
S##
#
##
#
S
## #
S#
S####
#
#
S
SS# #
#
S
#
S
#
#
S
#
#
S
#
#
S
S#
S
#
#
#
##
S#
S
#
#
#
S#
S####
S
#
S
S#
##
#
#
#
#
S#
#
#
S
S
S#
S
S
S
S
S#
#S
#
#
#
#
S
S
S
S#S
#
#
S#
##
S
#
S##
S
# ## ###
#
##
##
#
##
##
##
#
## #
Juvenile Coho
Salmon Density
0
0.01-0.50
0.51-1.0
1.01-2.0
2.01-10.0
#
##
##
5c. North Fork Alsea River, 2001
##
#
#
#
#
#
##
#
#
#
#
#
##
#
##
#
#
#
#
###
#
#
#
#
##
##
###
#
#
#
###
##
#####
##
#
5d. North Fork Alsea River, 2002
# ##
S
# ## #
##
##
#
#
#
#
S# #####
#
##
##
#
#
#
#
##
#
#
#
#
#
S
#
#
S
#
#
###
##
#
#
#
#
S
#
#
#
#
#######
#
#
#
S#
#
#
S
#####
#
#
###S#
#
S
S
S
##
#
#####
S
#
#
#
#
##
#
S
#
#
#
S
S
#
S
S
#
S
##
#
#
#
#
#
#
#
S
S
S
S
S
S
#
#
#
#
#
#
#
#
#
S
S
#
S
S
#
S
S
#
##
#
#
#
#
#
#
#
S
#
#
S##
#
#
S
# ##
S#
S#
S ##
S
##
S
S
#
S#
S#
#
S#
#
S
#
#
#
#
#
###
#
######
#
#
#
#### #
#
#
#
#
S
#
#
#
S#
#
S
#
#
##
S#
S#
S
##
#
S
#
#
S
##
#
S
#
S
#
##
#
S#
###
#
S
#
#
S
S
#
# #
#
#
S
### # ####
#
S
#
#
S
# ##
## #
S
S ###
#
S
S#
#
S
##
###
S
#
#
#
S
S
#
#
#
#
#
##
#
S
#
#
S
##
#
#
#
S
S
#
#
#
#
#
#
#
#
##
#
#
##
S##
#
#
#
S
S
#
#
#
S
S
S
S#
S
#
S
S# ##
#
#
#
#
S#
S#
S
S
S
S
S
#
#
#
S
#
S
#S
S
#
S
#
#
S
#
#
#
S
S
#
S
###
##
###
#
S
#
#
##
#
#
#
###
#
#
#
## #
###
#
###
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
# ##
#
#
S
#S
#
S S
##
#
##
####
#
#
#
#
###
S
#
S
#
S
#
S
#
S
#
#
#
#
#
S
#
#
#
#
#
#
Flitcroft
Chapter 3
51
Appendix 3.3. (continued)
6a. South Fork Alsea River, 1998
##
#
###
## ##
#
#
#
#
#
#
#
#
#
#
#
##
#
### #
#
#
#
#
S
#
#
#
#
# #
S
#
#
#
##
S
#
S
#
#
#
#
S ##
S#
#
S
#
#
S
### ### ##
#
S
#
#
S
#
#
#S#####
S
#
S
#
#
S
#
#
##
#
S
#
#
#
#
S
S
#
S
#
#
S
S
#
S
#
S
#
#
S
#
#
S
##
#
##
#
S
##
#
S
#
#
#
S
#
# #
#
S
S
#
#
S
##
#
#
##
#
S
#
#
#
S
#
S
#
S
#
#
#
S
#
S
S
#
S
#
#
S
S
##
#
S
#
#
S
S
S
#
S
#
S
#
#
S
S#
#
S
##
#
#
S
#
#
S#
S
#
S#
#
S#
S##
#
#
#
##
#
Juvenile Coho
Salmon Density
0
#
S#
S
#
S#
#
S
S
#
#
S#
#
S#
S#
S
#
0.01-0.50
0.51-1.0
#
#
#
#
#
##
1.01-2.0
2.01-10.0
6c. South Fork Alsea River, 2001
#
S
###
###
S
S
#
S#
#
S
#
S
#
S
S#
#
S
#
S#######
#
S
#
S##
#
#
#
#
S
#
#
S
S
#
S
#
#
S
#
#
S
#
S
#
#
S
S
# ##
#
S
SS
#
S
#
#
S
S
#
S
#
#
#
S
#
S
#
S
S##
#
S
S##
S##
S#
#
S
#
##
S#
#
S#
#
S#
S
#
S #
S
#
#
S
S
#
S
#
#
#
S
#
S
#
S
#
#
#
#
S ##
SS
#
#
S
S
#
#
#
S
S
S#
#
#
S#
S
#
#
S#
#
#
S
#
S
S##
#
#
#
S#
S
#
S
S
SS
S
##
#
S#
S#
#
S
#
S#S
#
#
S
#
#
S
#
S
#
#
S
#
S#
S
#
# #
S
S
#
S
#
S
S#
#
#
S ##
#
S
#
#
S
#
#
###
S
#
#
S
#
#
S
#
#
#
S
#
#
##
#
#
S
#
S#
#
#
#
#
#
#
#
#
#
#
##
6d. South Fork Alsea River, 2002
S#
##
#
S
#
S#
S
#
S ##
#
##
#
#
S
#
S
#
#
S
#
#
S
S
#
#
#
S
S#
#
#
S#
S######
S
#
S
#
## #
#
S#
S
#
#
#
S
#
S
#
#
S
S
###
#
S
#
S
#
#S##
S
#
S
#
#
S
#
S
#
#
S##
S
#
S#
S
#
#
##
S
S#
S
S
#
#
S#
S#
#
S
S
S#
#
#
SS#
S
#
S
S
#
S
#
S ###
S
#
S#
S
#
#
S
#
#
#
S
S
S
S#
#
#
#
S
S#
S
#
S
S#
#
#
S
S
S#
#
S
#
#
S#
S
#
S
#
#
S#
S
#
S
#
#
S
#
#
S
S
S
#
#
S#
#
#
S
S#
S
##
S
#
#
#
S#
S
S#
#
#
#
S
S
#
#
S#
#
S
S
#
S#
#
S
S
S#
#
#
S
S
#
S #
#
S
#
#
#
#
S#
S
#
#
S
#
S
#
#
S
#
S
#
#
#
S
##
S
##
#
S
##
#
S#
S
S
#
S
#
#
S#
#
# #
#
S
###
#
S#
##
#
#
#
#
#
#
#
#
#
#
##
#
#
#
#
#
#
#
#
#
S
#
S
#
S#
S
Flitcroft
Chapter 3
52
Appendix 3.3. (continued)
7b. Cedar Creek, 1999
Juvenile Coho
Salmon Density
0
0.01-0.50
#
#
S#
#
S
#
S
##
S
S
#
S
0.51-1.0
1.01-2.0
2.01-10.0
#
#
S
#
S
#
#
7c. Cedar Creek, 2001
7d. Cedar Creek, 2002
#
S
S
##
S
#
S
#
S#
#
S
#
S
#
S
#
S#
S###
#
S#
#
S
##
S
#
S
#
S
#
S
#
S###
S
#
#
SS
##
#
S
S#
S##
#
S #
#
#
#
S
S
#
S
#
S
#
S
#
S
#
S
S
#
S#
#
S
#
S
#
S
#
S
#
S
S#
#
S#
SS
#
S
#
S
#
#
S#
S
S#
# #
S####
#
S
#
S
#
S
#
#
S
S
#
S
#
S
#
#
S
#
#
S
S#
#
#
S
S
#
S
#
S
#
#
S
#
S#
#
S
S
#
S#
S###
########
#
#
#
#
S
#
#
## # ##
#
S
#
S#
S
##
#
#
#
S#
S
S##
#
#
S
#
S#
#
#
S
#
S#
S#
S
#
S
#
S#
S#
S
#
S
S#
#
S
#
#
S#
S
S
#
S
#
S #
####
##
S#
S#
S#S
S
#
S#
#
S
#
#
S
S
#
S#
#
#
#
S#
#
#
#
S
#
#
S
#
S
#
#
S
Flitcroft
Chapter 3
53
Appendix 3.3. (continued)
8b. Rock Creek, 1999
8a. Rock Creek, 1998
#
#
#
#
# #####
##### ### #
#
#
#
S
#
#
#
##
#
#
#
#
#
#
#
#
#
##
#
S
#
#
#
#
S
# #
#
#
#
S
#
S
#
S
# # # #
#
#
S
#
# #
#
##
#
S# ##
SS
##
S
#
####
S
#
#
#
S#
S
#
#
S #####
#
##
#
#
#
S
S
S
S#
S
S
#
#
#
#
S##
S
#
S#
S#
S
S
#
#
#
#
S
# #
##
#
S#
S
S
#
#
S
#
#
S
S#
S
#
S
##
#
S
#
#
#
S
#
S
##
S#
##
#
#
#
S
#
#
#
#
#
#
S
S
#
#
S
#
#
#
S
#
#
#
#
#
#
#
##
#
####
#
#
S#
#
S
#
#
#
#
#
##
#
#
S
#
#
#
#
#
S
##
#
S
# #
#
#
#
#
#
S
#
## #
#
#
S
#
#
S
#
#
S
#
# #####
# ##
#
S
#
S#
#
#
S
#
S
#
#
S
#
S#
##
#
S
#
S
S#
#
S
S
#
#
S
#
S
# #
S#
S##
S
S#
#
#
S#
S#
S#######
S
S#
#
S
S#
#
S
S#
## #
S
#
#
S#
#
#
#
#
#
#
#
#
#
#
S
##
S
#
S
#
S#
#
S#
#
#
#
#
#
#
S
#
#
#
#
#
#
S
#
#
S
S
S
#
#
#
#
#
#
#
#
S
S
S
#
S
S
S
#
S
#
S
S
##
#
S
#
#
#
S
#
#
S
S#
S
#
#
###
S#
S
S#
S
#
#
#
#
#
S
S
S
S
S
#
#
#
8c. Rock Creek, 2001
#
#
#
####
#
#
##
Juvenile Coho
Salmon Density
0
0.01-0.50
0.51-1.0
1.01-2.0
#
#
S
#
#
S
#
S
#
S#
#
#S
#
S
#
S
#
S
#
S
#
S
#
S
#
S
#
#
S
#
# #
S #
S
#
#
S
#
#
S
#
#
#
#
#
S
#
#
S
##
#
S
S#
#
#
#
#
S
S S ###
#
S#
S#
S#
S#
#
S#
S#
##
#
S
SS
S
S#
#
S
#
S
##
S#
# #
S
S
#
#
##
S
S#
#
S#
#
#
#
S
S
#
#
S
S
#
S
#
#
S
#
S#
S
#
#
#
#
##
#
#
2.01-10.0
##
###
#
##
#
##
#
#
#
#
#
#
#
#
#
#
# #
#
SS
#
#
#
##
###
###
#
#
#
#
#
S
S
##
#
S
####
#
#
#
#
#
S
#
#
S
#
S
#
S ##
#
#
S
##
#
#
S#
S
#
#
S
S
S
#
#
#
#
#
S
S
S
S
S
#
#
S#
#
S####
#
#
#
S#
S
S
S
S
S#
S
#
Flitcroft
Chapter 3
54
9b. Sams Creek, 1999
Juvenile Coho
Salmon Density
0
0.01-0.50
0.51-1.0
1.01-2.0
#
#
S# #
#
#
#
#
#
##
S#
#
#
S
#
S
#
#
S
S
#
#
#
S
S
#
S
S
#
S
#
#
S
S
#
S
S
S
#
#
#
#
#
S#
#
#
#
S
S
S
S
#
#
#
#
#
S
S##
S
#
S
S#
S
S
S#
S
S
#
#
#
#
S
S
#
S
S
S
S
#
#
#
#
S
#
S
#
S
S
#
S##
##
S
S
#
#
S
#
S
#
S
S
S
#
S
#
#
S
#
S
#
S
S
#
S
#
#
#
S
S
S#
S
#
S
#
#
#
#
#
S
S
S
S#
S##
#
S
S
#
##
#
#
#
S
#
S
#
#
S
#
#
S
#
#
S#
S####
#
#
S
#
#
S
S
#
S
S
#
#
S
##
###
#
S
#
S
S
#
#
#
S##
#
#
S
S
#
S
S#
#
S##
#
S#
S
S
#
S#
S
#
S
#
#
SS
S
#
S
# #
#
S##
#
S
#
#
#
#
2.01-10.0
9c. Sams Creek, 2001
#
#
####
#
##
##
#
#
#
#
#
##
#
#
#
#
#
#
#
#
##
#
#
##
##
#
#
#
# #
##
#
#
S
#
#
#
#
#
S
#
##
S
#
#
S
#
#
###S#
#
#
9d. Sams Creek, 2002
##
#S
S
S#
S#
S#
S
#
#
#
#
#
S
#
S
### #
#
S
#
#
S#
#
#
##
###
S
#
#
#
S#
#
#
S## S
S
S
####
# #
S #
##
S#
#
S#
#
#
#
#
#
#
#
#
##
S
##
#
#
SS
#
S
###
#
#
#
S
S
#
#
S#
#
#
#
###
#
##
##
# ##
#
#
#
#
###
#
#
###
#
#
#
#
#
#
#
#
#
#
#
#
####
#
#
#
#
#
#
#
##
#
#
##
###
###
##
##
##
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
# ###
#
##
#
#
#
#
#
#
#
S#
##
##
#S
S
S
S
S#
S#
S#
S
#
S
#
#
S
#
#
S#
S#
#
S
S# ##
#
#
S
#
#
#
S##
##
S
#
#
#
###
#
##
# #
# ####
#
#
#
S
## # #
## S#
#
#
S
#
#
#
S#
S#
#
#
#S#
#
##
#
#
#
#
#
#
#
S
#
#
#
#
#
S
#
##
#
#
## #
#
S
S
#
##
#
#
##
#
#
## ##
S
S
#
S
##
#
#
## ##
#####
###
## ##
##
# #
#
##
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
###
# #
#
#
#
#
#
##
#######
#
#
#
#
#
##
##
##
#
#
#
#
#
#
#
#
#
#
#
#
###
#
#
#
#
#
##
Rebecca L. Flitcroft. 2007. Regions to streams: Spatial and temporal Variation in stream occupancy patterns of coho salmon (Oncorhynchus
kisutch) on the Oregon Coast. Ph.D. dissertation. Department of Fisheries and Wildlife, Oregon State University, Corvallis, OR.
CHAPTER 4
Multiple Spatial Scales in an Analysis of Two Life History Stages of Coho Salmon
(Oncorhynchus kisutch): A Hierarchical Bayesian Approach
Rebecca L. Flitcroft
Department of Fisheries and Wildlife
Oregon State University
Corvallis, Oregon
Abstract
Life history diversity over broad spatial extents characterizes populations of anadromous salmonids in the Pacific
Northwest. The life history of coho salmon, Oncorhynchus kisutch, is a complex example of adaptation in response
to diverse habitats in a dynamic environment. It is possible that ecological processes acting at different spatial scales
influence the abundance and distribution of coho salmon. Adult distribution reflects dispersal patterns adapted
to maintain salmonid populations as large scale disturbances render entire basins suitable or unsuitable over time.
Meanwhile, juvenile distribution reflects adaptation to local conditions as individuals seek habitats for a year of freshwater
occupancy. The usefulness of incorporating multiple spatial scales in models of abundance for each life history stage
was explored in this study. Hierarchical modeling of two spatial scales for each life history stage was completed using
Bayesian methods. Results indicated that adult coho salmon abundance could be explained using two-level hierarchical
models that incorporated the subbasin variable, percent large trees in a 100-m riparian buffer, modeled with the basin
scale variable, mean annual precipitation. This contrasts with juvenile distributions that were adequately represented by
a two-level hierarchical model that included three site scale variables, percent sand, stream order, and network distance
to spawning habitat, hierarchically linked to non-specific subbasin scale variation. This suggests that effective recovery
strategies for endangered salmonids may require targeting habitat features at different spatial scales that are relevant to
individual life history stages.
Introduction
The spatial scale at which scientists analyze the ecological
world greatly influences a studys’ result (Wiens 1989) which
means that choosing the proper spatial scale for analysis
is important and may be challenging. Small spatial scales
are readily accessible to researchers (Matthews and MarshMatthews 2003) even when the species under investigation may
also be described at large spatial scales. Unfortunately, small
spatial scales may not capture the diversity of environmental
conditions under which individual species evolved (Pacific
Fisheries Resource Conservation Council 1999). Also, the
same ecological process may manifest differently across spatial
scales (Wiens 1989). Therefore, studies that examine small
scales may not accurately address questions of species needs
or ecological processes that are important for larger scale
management or restoration efforts.
It may also be important to incorporate multiple spatial
scales in order to explore potential synergisms between scales
when addressing ecological phenomena (Lowe et al. 2006).
Because ecological processes do not work in isolation, there
may be interaction among processes, thereby introducing
interactions across scales. Multiple scales included in an
analysis may better represent interactions between geomorphic
and ecological processes. Poff and Huryn (1998) found that
multiple scales of processes interact in the determination of
freshwater production of Atlantic salmon (Salmo salar). They
concluded that a multi-scale perspective was informative in
determining conservation, restoration and management
actions. Such a multi-scale perspective could be useful in
approaching analysis of other wide ranging anadromous fish.
Some aquatic ecologists are beginning to recognize the
importance of analysis that encompasses entire stream systems
(Fausch et al. 2002). For example, Torgersen et al. (2004) used
continuous surveys of coastal cutthroat trout (O. clarki clarki)
to describe differences in patterns of juvenile occupancy with
underlying geomorphic conditions. They found that analysis
of continuous survey data could be used to precisely correlate
Flitcroft
Chapter 4
patterns of juvenile occupancy at three spatial scales that
corresponded with local stream conditions and the network
structure of tributaries in the stream system. It is possible that
analysis encompassing spatial scales ranging from individual
sites to regions may contribute important information
regarding abundance, distribution, and life history needs
that is necessary to understand the adaptation and survival of
anadromous fish species.
Anadromous fish species exemplify a life strategy that
encompasses multiple spatial scales and extents at different
life stages (Armstrong 2005, Bradford et al. 1997, Myers et
al. 1997). For example, Pacific salmon (Oncorhynchus spp.)
may migrate thousands of miles in the ocean. The majority
of mature individuals in a population return to their natal
freshwater streams to spawn, while some may stray into
neighboring streams or entirely different basins or regions
(Cooper and Mangel 1999, Labelle 1992). Therefore, the
spatial extent for adult anadromous salmon movement includes
potentially extensive tracts of ocean as well as freshwater
areas. The diversity of adult behavior that involves homing
or straying allows for the distribution of genetic material and
opportunities to explore and recolonize habitat over large areas
and ultimately contributes to population persistence (Cooper
and Mangel 1999).
Depending on the species, juvenile anadromous salmonids
may use the stream system for up to two years before migrating
to the ocean to mature. For juveniles that overwinter in
streams, the spatial extent occupied in the stream network by
the juvenile life stage is comparable to that of adults. However,
unlike adults, juveniles do not migrate between basins
through the ocean as spawning adults may do when they
home or stray. Therefore, the spatial extent of movement of
juveniles is smaller than that of adults, with juveniles restricted
by mobility and life history needs to the basin of their birth.
Within their natal stream system, juveniles display a variety of
movement behaviors with some moving only small distances
while others move among tributaries (Murray and Rosenau
1989, Quinn 2005). Variety of movement behaviors allows
different individuals to take advantage of the habitats of the
stream environment (Nielsen 1992).
In this study I explore spatial scale, stream network variables,
instream habitat conditions and regional habitat variation in
assessing the abundance of adult and juvenile coho salmon
(O. kisutch) in western Oregon. Coho salmon are a natural
choice for an investigation of the relationship between spatial
scales and life history stage because of the extensive research
on freshwater habitat preferences for adult and juvenile fishes.
In addition, monitoring efforts have made available data
documenting entire populations of coho salmon at juvenile
and adult life stages. In the course of their lives, coho salmon
may migrate thousands of miles from their birth streams to
56
the ocean and back again. The spatial scale of movement and
dispersal of adult coho salmon encompasses large regions of
freshwater through ocean migration and subsequent homing
or straying to spawn (Labelle 1992, Li et al. 1995). The spatial
extent of freshwater movement by juvenile coho salmon
contrasts with adults. Juveniles occupy multiple habitat types
in fresh water and may move among tributaries in their birth
streams (Kahler et al. 2001, Bolton et al. 2002) but do not
tend to move between large basins as adults may when they
return to freshwater to spawn.
An assumption of this work is that adults define the
distribution of populations of coho salmon both within stream
systems and among stream basins. I will refer to adult coho
salmon freshwater distribution as the topology of dispersal. The
term dispersal implies movement and change while topology
(Flitcroft 2007b) refers to the overall distribution of adults in
the stream network. Another assumption is that juveniles do
not move beyond their natal basin and the pattern of their
distribution reflects local survivorship and the spawning
location of their parents (Flitcroft 2007b). I will refer to the
pattern of juvenile coho salmon freshwater stream use as the
topology of juvenile occupancy. The term occupancy reflects
the year of juvenile freshwater residence in the stream network.
Three spatial scales (site, subbasin, and basin) (Figure 4.1)
were chosen for analysis and are meant to represent different
resolutions of habitat and fish movement. The site scale is the
smallest, and includes local habitat and the proximity among
habitats in the stream network. The subbasin (5th-field
Hydrologic Unit ranging in size between 11,179 and 33,225
hectares) represents a drainage network small enough to allow
adults and juveniles to move within and among networks.
The basin (4th-field Hydrologic Unit ranging in size between
33,787 and 291,746 hectares) represents a drainage network
large enough to contain all potential juvenile movement, while
allowing adults to choose between basins when returning to
spawn.
In this research, the usefulness of including multiple
spatial scales in assessments of the abundance of the juvenile
and adult life history stages of coho salmon was explored.
First, I hypothesize that the spatial scales appropriate for
understanding adult density are subbasins and basins. When
the metapopulation organization of the fish (Cooper and
Mangel 1999), the large scale of movement, and the fact that
adult salmon are the colonizers and recolonizers of basins are
considered (Labelle 1992), I speculate that the scale of the
ecological process of adult coho salmon dispersal will be basins
coupled with the moderate scale of the subbasin. I explored
this hypothesis using datasets that encompass the ocean
draining streams in coastal Oregon (Figure 4.2).
Secondly, I hypothesize that the spatial scales of sites and
subbasins will be important in understanding juvenile density.
Flitcroft
Figure 4.1. Three spatial scales used in
hierarchical Bayesian analysis of adult and
juvenile coho salmon. Basin and subbasin
scales were used to hierarchically model adult
coho salmon and subbasin and site scales
were used to hierarchically model juveniles.
Figure 4.2. Oregon’s coastal region
encompassed the 5th field and 4th field
hydrologic units used in a multi-scale
hierarchical Bayesian analysis.
Chapter 4
57
Flitcroft
Chapter 4
This reflects the limited movement of individual juveniles in
their year of freshwater occupancy (Bell 2001, Bramblett et
al. 2002, Quinn and Peterson 1996). I explore this hypothesis
in six subbasins of the Alsea (Upper Drift Creek, Five Rivers
including Lobster Creek, and Upper Alsea River) and
Siletz (Middle Fork Siletz and Lower Siletz River and Rock
Creek) river systems in Oregon’s mid-coast region in which
comprehensive survey information for juveniles was available
(Figure 4.2).
Hierarchical Bayesian models that incorporated spatial
variables at scales appropriate for local sites, subbasins and
basins were used for analysis. A hierarchical Bayesian approach
provided flexibility in the hierarchical structure of the models
(Browne and Draper 2006, Carlin et al. 2006, Gelman et
al. 2004). Hierarchical Bayesian methods have been used to
explore a variety of ecological questions including monitoring
of harbor seals over time (VerHoef and Frost 2003), the
incorporation of spatial variation in population modeling
(Gelfand et al. 2006), and modeling multi-scale relationships
(Monlean et al. 2002). Use of hierarchical Bayesian models is
rare in fish ecology literature possibly reflecting the novelty of
incorporating multiple spatial scales in analysis. Two examples
include Wyatt (2002) who used Bayesian hierarchical models
to map fish populations in a river network and Harley and
Myers (2001) who modeled trawl survey effectiveness.
However, applications of other Bayesian methods such as the
Kalman Filter are not uncommon in fish ecology (Lee 2000,
Lee and Rieman 1997, Marcot et al. 2001).
Methods
Study Area
Coastal basins in Oregon were used in this study (Figure
4.2). Oregon’s coastal region is demarcated to the west by the
Pacific Ocean and to the east by the Coast Range Mountains.
The coastal region is characterized by a mild climate with
precipitation heaviest in the winter months (Redmond and
Taylor 1997). The geology of Oregon’s coast ranges from
sedimentary to volcanic, with volcanic rock types found
towards the north. Lowland areas are fertile for agriculture and
coniferous forests are found throughout the region. Timber
harvest and fishing are strong economic enterprises in coastal
Oregon and are also culturally important (Lichatowich 1999).
Dtasets and Variables
The exploration of multiple spatial scales requires the
acquisition of data describing adult and juvenile run sizes,
stream characteristics such as instream habitat and riparian
vegetation, geology and rainfall. After a preliminary literature
review, I identified six categories of stream characteristics
58
as important: habitat structure, shape/sinuosity, network
relationships, landscape conditions, lithology and rainfall.
Within each category, variables were identified that represent
each scale (Table 4.1). Datasets were compiled that contained
the variables of interest (Table 4.2). Averaging variables over
space is a distortion of information (Gotway and Young 2002).
This consideration guided the selection of variables that were
meant to accurately represent each scale. For example, a
variable associated with the large spatial extent of a basin would
need to characterize the entire area, compared to a variable at
the small spatial extent of study site which would reflect local
conditions such as the amount and type of substrates available
at the site. This is particularly important when using remotely
sensed data that is collected based on pixel sizes that may range
from 1 meter to 100 meters in size. A large pixel size such as
100 meters will contain coarser information than the 1-meter
pixel. Therefore, a 100-meter pixel may not be appropriate for
site specific habitat characterization, but may be adequate for
basin scale descriptions.
Juvenile coho salmon survey counts were conducted by the
Mid-Coast Watershed Council. Every fifth pool was snorkeled
and all fish were counted from the stream mouth to the
headwaters, including all tributaries, during the summer of
2002 (June through September). A summer juvenile survey
is meant to capture a snapshot of coho salmon distribution
during the time of year when the fish are assumed to move
the least (Nickelson et al. 1992). Density was calculated by
dividing fish counts by the length of the pool unit, and was the
measure used for modeling.
Information about spawning coho salmon came from
the Oregon Department of Fish and Wildlife’s (ODFW)
Spawning Survey program for spawners in 2001 (spawning
season of October 2001 through February 2002) and 2002
(spawning season of October 2002 through February 2003).
Field survey crews repeatedly visited survey sites (selected using
a spatially random design) in the study basins through the
spawning season as part of the effort to monitor spawning coho
salmon across the entire coast (Jacobs and Nickelson 1989,
Jacobs et al. 2002). The salmon spawning survey sites were
modified using a GIS with a protocol developed by researchers
associated with the Environmental Protection Agency’s STAR
Grant program (R. Smith 2007 personal communication).
This allowed for the interpolation of the spawning survey sites
across the landscape using a grid of points (R. Smith 2007
personal communication). The final product was a landscape
interpolation that displayed the estimated number of spawners
across the entire coast for each year of spawning data. The
mean number of spawners at the subbasin scale was used for
modeling.
Instream fish habitat survey information came from the
Oregon Department of Fish and Wildlife’s Aquatic Inventories
Program (ODFW-AIP). Field habitat survey crews walk from
Flitcroft
Chapter 4
59
Table 4.1. Variables that represent different spatial scale were used in multi-scale hierarchical Bayesian models of the adult and juvenile life
history stages of coho salmon. *Variable was standardized ((X-(mean))/standard deviation).
Scale
Habitat Structure
Network
Relationships
Shape
Sinuosity
Landscape
Condition
Lithology
Percent gravel
Percent sand
Percent bedrock
Rainfall
Site
*No. pieces of wood
*Boulder count
Slope
*Distance to spawning
*Distance to rearing
*Distance to summer
*Stream order
---
Subbasin
(5th-Field
HUC)
Percent lrg trees in
100-m riparian buffer
Percent subbasin in
valley class
*Number of
tribs
*Area
Percent subbasin km in
high intrinsic potential
class
Percent subbasin Percent sedimentary
in conifer veg
geology
class
Percent volcanic
geology
*Mean annual
precipitation
Basin
(4th-Field
HUC)
Percent lrg trees in
100-m riparian buffer
Percent basin in valley
class
*Number of
tribs
*Area
Percent basin km in high Percent basin in Percent sedimentary
intrinsic potential class conifer veg class geology
Percent volcanic
geology
*Mean annual
precipitation
---
Table 4.2. Datasets that were assembled to derive variables that could be used to represent multiple spatial scales.
Dataset
Spawning
Salmon
Source
Oregon Dept. of Fish
and Wildlife Salmon
Spawning Survey
Program
Juvenile
Mid Coast Basin
Snorkel Counts Council
Available Area
Variables
How Data Were Collected
Coastal draining streams adult spawning coho
of Western Oregon.
density
The sample sites were identified using a stratified random sample design (Jacobs
and Nickelson 1998) and were used in the long term monitoring of trends in
spawning populations in Western Oregon (Jacobs et al. 2000, Jacobs et al. 2001,
Jacobs et al. 2002). Data collected at these sites included the number of coho
redds and adults on the spawning grounds. Sample sites were revisited multiple
times throughout the spawning season. The salmon counted are called naturally
spawning adults (Jacobs et al. 2002).
Alsea River and Siletz
River basins
juvenile coho density
This dataset was a systematic sample of every fifth pool from the mouth of the
river to the upper extent of juvenile coho distribution. The upper extent was
determined as the point at which 5 consecutively sampled pools held no fish
(Mills 2003). The entire stream network including tributaries bearing coho were
surveyed and the data recorded.
Coastal draining streams
of Western Oregon from
Astoria in the north
to Cape Blanco in the
south
active channel width,
gradient, stream order,
percent basin in valley
classification, percent
basin km in high
intrinsic potential
classification
Stream networks and geomorphic characteristics were modeled from 10 meter
digital elevation models (DEM’s) (Miller 2003). Supplemental information such
as precipitation records were used to characterize water flow.
Coastal draining basins
of Western Oregon from
Astoria in the north
to Cape Blanco in the
south
percent basin in conifer
vegetation class, percent
large trees in 100m
riparian buffer
This layer was created using a Gradient Nearest Neighbor method that
integrated vegetation measurements from regional grids of field plots, mapped
environmental data, and Landsat TM imagery (Ohmann and Gregory 1996).
Scattered streams
throughout Oregon
percent gravel substrate
in unit, percent sand
substrate in unit,
percent bedrock
substrate in unit
Continuous instream habitat survey information was gathered by field crews
that walked the entire stream from mouth to headwaters. The stream is broken
down into habitat units such as pools, riffles and glides. Measurements such as
water depth, substrate composition, boulder counts and wood were taken for
each habitat unit (Moore et al. 1997).
mean annual
precipitation for the
basin
Rainfall records were compiled by the PRISM group at Oregon State University
through the Oregon Climate Service. Rainfall records between 1971 and 2000
were used in an assessment process that incorporated spatial effects of rain
shadows and coastal effects (Daly and Johnson 1998).
percent basin with
sedimentary geology,
percent basin with
volcanic geology
Geology was digitized from the 1:500,000 Oregon Geologic Map, which was
created by the USGS.
Stream Lines
USDA Forest Service
Coastal Landscape
Analysis and Modeling
Study
Vegetation
USDA Forest Service
Coastal Landscape
Analysis and Modeling
Study
Instream
Habitat
Oregon Dept. of Fish
and Wildlife Aquatic
Inventories Project
Precipitation
PRISM group at Oregon
State University All of Oregon
provided through the
Oregon Climate Service
Geology
US Geological Survey
All of Oregon
Flitcroft
Chapter 4
the stream mouth to the headwaters during the summer,
taking measurements of the physical stream environment
and identifying discrete habitat units such as pools, riffles and
glides (Moore et al. 1997). Field surveys from 1997 through
2002 were pieced together to encompass the subbasins of
interest for this study.
Geology classifications were acquired from a dataset
created by the United States Geological Survey (USGS) that
contained digitized geology information at a 1:500 000 scale.
Mean annual precipitation that incorporated the spatial effects
of rain shadows and coastal influences was compiled by the
PRISM (Parameter-elevation Regressions on Independent
Slopes Model) group at Oregon State University through the
Oregon Climate Service (Daly et al. 1999).
Stream linework and stream network characteristics were
modeled by the Coastal Landscape Analysis and Modeling
Study (CLAMS) based on 1:10-m digital elevation models
(DEMs) (Miller 2003). Vegetation was also mapped by
CLAMS based on aerial photographs and remotely sensed
imagery (Ohmann and Gregory 2002).
As with any project that seeks to use datasets collected by
different groups for different purposes, a significant challenge
was identifying consistent and compatible data from different
sources that could then be compiled for this study. I chose
to develop a geographic information system (GIS) to
produce spatially accurate map layers. First, I georeferenced
ODFW habitat data and the juvenile snorkel survey counts
to the stream linework generated by CLAMS. The dynamic
segmentation protocol in ESRI’s ArcInfo software was used to
attach information to the linework (Tables 4.1 and 4.2).
At the scale of a site, the location of juvenile coho salmon
snorkel surveys was used as the spatial base. Information
was extracted from the ODFW habitat surveys and CLAMS
stream layers using a 100-m buffer around each site. At the
scales of subbasin and basin, variables of interest (i.e., geology,
vegetation, precipitation) were calculated as the mean value
over the area of interest. Riparian information was derived
Table 4.3. The number of 5th-field HUCs within
4th-field HUCs and the number of sites sampled
in each 5th-field HUC.
60
from the vegetation layer by buffering (100 m) fish bearing
streams and calculating the percent of the buffer covered by
large trees.
The number of observations varies at each spatial scale. At
the site scale there were a total of 742 observations. At the
subbasin scale there were 59 observations, and at the basin
scale there were 10 observations. The number of observations
for juvenile models exploring site and subbasin variation
include 742 sites and 6 subbasins. The number of observations
for adult models using subbasin and basin scales includes 59
subbasins and 10 basins (Table 4.3). All explanatory variables
that were not a percentage were standardized in order to
facilitate convergence of simulation models ((X-(mean))/
standard deviation).
Bayesian Methods and Models
In my first hypothesis I propose that spatial scales
appropriate for understanding adult coho salmon abundance
are subbasin and basin. In my second hypothesis, I propose
that the spatial scales appropriate for understanding juvenile
coho salmon density are site and subbasin. The hypotheses
for both juvenile and adult coho salmon explored the use of
multiple spatial scales for analysis and description. In order
to explore these hypothesis, I used Bayesian hierarchical
modeling, which allows the use of information from each
basin for adults, and each subbasin for juvenile coho salmon,
without the restriction that effects or variation within basins
or subbasins must remain constant. While frequentist models
allow this as well, Bayesian methods are more intuitive and
easier to interpret (Carlin et al. 2006).
In this application, the different levels of the Bayesian
hierarchical models correspond with spatial scales of interest.
Models for adult coho salmon include two hierarchical levels
that correspond to subbasin and basin scales (Figure 4.3a).
Models for juvenile coho salmon also include two hierarchical
levels that correspond to site and subbasin scales (Figure
Adult Spawner Dataset
4th-Field
HUC
17100201
17100202
17100203
17100204
17100205
17100206
17100207
17100303
17100304
17100305
Juvenile Dataset
HUC Area # of 5th-Field
(hectares)
HUC’s
35426
221214
249861
196371
178566
200744
33787
391746
190672
273714
1
6
9
10
7
8
1
7
4
6
5th-Field
HUC
HUC Area
(hectares)
Number
of Sites
1710020405
1710020406
1710020407
1710020501
1710020502
1710020503
16763
11179
32415
33225
30903
17925
123
55
178
131
85
182
Flitcroft
Chapter 4
61
Figure 4.3. Hierarchical Bayesian model structure for adult (a) and juvenile (b) coho salmon.
a. Structure of hierarchical
model describing adult
spawner abundance
b. Structure of
hierarchical model
describing juvenile coho
density
4.3b). Models were evaluated using the Bayesian Information
Criterion (BIC), which captures the residual variance and a
penalty term.
The penalty term combines the number of parameters and
the natural log of the number of samples, in an attempt to
compensate for the ability of large sample sizes to identify
statistically significant but unimportant coefficients (Ramsey
and Schaefer 2002). By nesting the hierarchical structure of
the models using spatial scale I hoped to control for spatial
autocorrelation (Gelfand et al. 2006).
The program WinBugs 1.4 (Spiegelhalter et al. 2003) was
used to fit the Bayesian models. Unlike modeling software
available for frequentist statistics, WinBugs requires that
each model be written and run individually because there is
no automated stepwise procedure available. A small group
of biologically meaningful models were written and run;
sixteen models of adults in 2001 and 2002 and ten models
of juveniles for 2002 (Table 4.4 and Table 4.5 respectively).
Further models were written and run by refining models with
the lowest BIC by reducing variables. The parameters selected
for reduction were chosen because they had limited utility in
estimating the mean of spawner abundance or juvenile density.
The usefulness of a variable was determined by examining the
parameter’s estimate and posterior interval. Posterior intervals
that include zero imply that the variable may not be useful.
By assuming that the response variables of juvenile density
or adult spawner abundance follow normal distributions, I
fit Bayesian hierarchical models with normal likelihood and
either normal or non-informative priors. I am assuming
that the ecological data fits a normal` distribution. For some
variables I chose to use non-informative priors – a conservative
choice because I am not familiar with other studies that could
give guidance to the choice of prior values. As an example, I
present adult spawner model number 5b (Table 4.4).
Example: Adult Spawner Model Number 5b. Let j = 1,
2,…., 10 index the 10 basins of the Oregon coast in which this
model pertains. Let i = 1,….,nj index the number of subbasins
present in each basin j.
Yij ~ N (θij,1/σ2j)
*Basin level
θij = B0j + B1j(% large trees in riparian zone)ij
B0j ~ Normal (0, 0.001)
B1j ~ Normal (μ1j, 1/σ2)
μ1j = B2 + B3(mean annual precipitation)j
*Subbasin level
1/σ j = gamma (0.01, 0.01)
1/σ2 = gamma (0.01, 0.01)
2
In this model, I am seeking to identify Yij (the abundance of
adult spawners based on subbasin i in basin j) conditional on θ ij
with a precision of 1/ σ2j. The ij are conditionally independent
and drawn from some population of θ. In this model, θij is
a linear function of the percent large conifers in the riparian
zone of the subbasin. The average slope of this relationship is
a linear function of the mean annual precipitation at the basin
level.
Flitcroft
Chapter 4
62
Table 4.4. Bayesian hierarchical models for adult coho salmon on the Oregon coast. For each model, subbasin-scale variables that are
hierarchically connected to basin-scale variables are listed next to one another in the table columns. The Bayesian Information Criterion or
BIC value allows for a comparison of model effectiveness by capturing the residual variance and a penalty term. The model with the lowest
BIC value is considered to be the best fitting model.
Model
Number Subbasin-scale Variables
Basin-scale Variables
BIC, 2001 BIC, 2002
1
Percent large trees in 100-m riparian buffer;
Area;
Percent subbasin in high intrinsic potential;
Percent subbasin in sedimentary geology;
Percent subbasin in volcanic geology;
Mean annual precipitation
Percent basin in conifer vegetation class;
Area;
—
—
—
Percent basin in sedimentary geology
801.8
853.469
2
Percent large trees in 100-m riparian buffer;
Percent subbasin in valley classification;
Number of tributaries;
Area;
Percent subbasin in high intrinsic potential;
Percent subbasin in sedimentary geology;
Percent subbasin in volcanic geology;
Mean annual precipitation
Percent basin in conifer vegetation class;
Percent basin in conifer vegetation class;
Area;
Area;
Percent basin in high intrinsic potential;
—
—
Percent basin in sedimentary geology
914.467
971.093
3
Percent subbasin in valley classification;
Percent subbasin in high intrinsic potential;
Percent subbasin in conifer vegetation class;
Percent subbasin in sedimentary geology;
Mean annual precipitation;
Percent subbasin in volcanic geology
Area;
Mean annual precipitation;
Percent basin in conifer vegetation class;
Mean annual precipitation;
—
Mean annual precipitation
811.938
877.450
4
Percent large trees in 100-m riparian buffer;
Percent subbasin in high intrinsic potential;
Percent subbasin in sedimentary geology;
Percent subbasin in volcanic geology
Mean annual precipitation;
Mean annual precipitation;
Percent basin in conifer vegetation class;
—
716.114
771.228
5
Percent large trees in 100-m riparian buffer;
Percent subbasin in high intrinsic potential
Mean annual precipitation;
Percent basin in sedimentary geology
616.312
686.807
5b
Percent large trees in 100-m riparian buffer
Mean annual precipitation
603.107
637.881
6
Percent large trees in 100-m riparian buffer;
Percent subbasin in sedimentary geology;
Percent subbasin in volcanic geology
—
Mean annual precipitation;
Mean annual precipitation
694.491
720.085
7
Percent large trees in 100-m riparian buffer;
Percent subbasin in high intrinsic potential;
Percent subbasin in sedimentary geology;
Percent subbasin in volcanic geology
Percent basin in conifer vegetation class;
Mean annual precipitation;
Mean annual precipitation;
Mean annual precipitation
726.729
785.104
8
Percent large trees in 100-m riparian buffer;
Percent subbasin in high intrinsic potential;
Percent subbasin in sedimentary geology
Percent basin in conifer vegetation class;
—
Mean annual precipitation
649.651
719.133
Flitcroft
Chapter 4
Table 4.4. (continued)
Model
Number Subbasin-scale Variables
Basin-scale Variables
63
BIC, 2001 BIC, 2002
9
Area;
Percent large trees in 100-m riparian buffer;
Percent subbasin in high intrinsic potential;
Percent subbasin in conifer vegetation class
—
Percent basin in volcanic geology;
Percent basin in conifer vegetation class;
—
703.840
757.154
10
Percent subbasin in high intrinsic potential;
Mean annual precipitation
Percent basin in sedimentary geology;
Area
677.200
700.701
11
Percent subbasin in high intrinsic potential;
Percent subbasin in conifer vegetation class;
Percent subbasin in sedimentary geology;
Number of tributaries
Mean annual precipitation;
Percent basin in high intrinsic potential;
Mean annual precipitation;
—
734.423
764.998
12
Area;
Percent large trees in 100-m riparian buffer;
Percent subbasin in conifer vegetation class;
Percent subbasin in sedimentary geology;
Percent subbasin in volcanic geology
—
Percent basin in valley classification;
Percent basin in high intrinsic potential;
—
—
746.984
793.538
13
Percent large trees in 100-m riparian buffer;
Percent subbasin in valley classification
Mean annual precipitation;
Percent basin in high intrinsic potential
611.470
694.305
14
Percent large trees in 100-m riparian buffer;
Mean annual precipitation
Percent basin in sedimentary geology;
Area
656.638
689.220
14b
Percent large trees in 100-m riparian buffer;
Mean annual precipitation
—
Area
644.078
674.628
15
Percent subbasin in high intrinsic potential;
Mean annual precipitation;
Percent subbasin in valley classification;
Number of tributaries
Percent basin in conifer vegetation class;
Area;
Mean annual precipitation;
Percent basin in valley classification
736.767
796.897
16
Percent large trees in 100-m riparian buffer;
Number of tributaries;
Percent subbasin in sedimentary geology;
Percent subbasin in volcanic geology;
Mean annual precipitation
Percent basin in sedimentary geology;
Percent basin in valley classification;
Mean annual precipitation;
Mean annual precipitation;
Number of tributaries
805.234
828.679
Flitcroft
Chapter 4
64
Table 4.5. Bayesian hierarchical models for juvenile coho salmon in subbasins of Oregon’s mid-coast. For each model, site-scale variables that
are hierarchically connected to subbasin-scale variables and are listed next to one another in the table columns. The Bayesian Information
Criterion or BIC value allows for a comparison of model effectiveness by capturing the residual variance and a penalty term. The model
with the lowest BIC value is considered to be the best fitting model.
Model
Number Site-scale Variables
Subbasin-scale Variables
BIC, 2002
1
Wood count;
Network distance to spawning habitat;
Network distance to rearing habitat;
Network distance to summer habitat;
Slope;
Percent sand in habitat unit;
Percent gravel in habitat unit;
Percent bedrock in habitat unit;
Boulder count
Percent subbasin in valley classification;
Percent subbasin in valley classification;
Number of tributaries;
Percent subbasin in high intrinsic potential;
Percent subbasin in sedimentary geology;
Percent subbasin in sedimentary geology;
Percent subbasin in sedimentary geology;
Percent subbasin in volcanic geology;
Percent subbasin in volcanic geology
3036.45
2
Wood count;
Network distance to rearing habitat
Percent subbasin in conifer vegetation class;
Number of tributaries
2635.68
3
Slope;
Network distance to rearing habitat;
Percent gravel in habitat unit
Percent subbasin in valley classification;
Number of tributaries;
Percent large trees in 100-m riparian buffer
2678.59
4
Wood count;
Slope;
Stream order;
Network distance to rearing habitat;
Percent bedrock in habitat unit
Percent large trees in 100-m riparian buffer;
Percent subbasin in valley classification;
Number of tributaries;
Percent subbasin in high intrinsic potential;
Percent subbasin in volcanic geology
2802.46
5
Network distance to spawning habitat;
Network distance to rearing habitat;
Network distance to summer habitat;
Stream order;
Slope
Percent subbasin in valley classification;
Number of tributaries;
Percent subbasin in high intrinsic potential;
Number of tributaries;
Percent of subbasin in high intrinsic potential
2790.70
6
Wood count;
Boulder count;
Slope;
Percent sand in habitat unit;
Percent gravel in habitat unit
Percent large trees in 100-m riparian buffer;
Percent subbasin in sedimentary geology;
Percent subbasin in sedimentary geology;
Percent subbasin in sedimentary geology;
Percent subbasin in sedimentary geology
2814.73
7
Wood count;
Network distance to rearing habitat
Percent large trees in 100-m riparian buffer;
Number of tributaries
2639.28
8
Stream order;
Network distance to spawning habitat;
Percent gravel in habitat unit
Number of tributaries;
Percent subbasin in sedimentary geology;
Percent subbasin in valley classification
2671.80
9
Percent sand in habitat unit;
Percent subbasin in sedimentary geology +
percent subbasin in volcanic geology;
—
—
2616.09
Stream order;
Network distance to spawning habitat
Flitcroft
Chapter 4
65
Table 4.5. (continued)
Model
Number Site-scale Variables
Subbasin-scale Variables
BIC, 2002
9b
Percent sand in habitat unit;
Stream order;
Network distance to spawning habitat
Percent subbasin in sedimentary geology;
—
—
2609.42
9c
Percent sand in habitat unit;
Stream order;
Network distance to spawning habitat
—
—
—
2581.41
10
Slope;
Network distance to spawning habitat;
Stream order
Percent subbasin in valley classification;
Percent subbasin in high intrinsic potential;
Number of tributaries
2669.43
Results
Adults
Based on BIC values, three of the top four models for
adults in 2001 and 2002 were the same (Table 4.6). These
three models (numbers 5, 13, and 14) were selected for
further assessment. In model number 5, one of the sets of
parameters was removed (percent subbasin in high intrinsic
potential at the subbasin scale connected with percent basin
in sedimentary geology at the basin scale). The resulting
model (5b) had the lowest BIC for 2001 and 2002 among all
models evaluated and included the subbasin variable, percent
large trees in the 100-m riparian buffer, connected with mean
annual precipitation at the basin scale. Model number 13 was
not altered because alteration would have led to a model the
same as 5b. Model 14 was modified by removing the basin
scale parameter percent basin in sedimentary geology. This
resulted in a model with a lower BIC than the original model,
but still higher than 5b (Table 4.6).
Aside from the identification of models with low BIC,
another result may be gleaned from an evaluation of the entire
group of adult models. Parameters that were important in the
explanation of the dependent variable are assessed by examining
the 95% posterior interval for the coefficient. If this interval
contains zero it suggests that the parameter is not useful. In
the adult hierarchical models, meaningful parameters that
describe basin level variation were uncommon. However, in the
models where basin level variables meaningfully contributed
to the model, the same combination of variables was visible
(Table 4.7). Mean annual precipitation and the number of
tributaries at the basin scale seemed to be connected with the
geology and percent of conifers in the riparian zone. These
relationships appeared to be important in modeling the mean
adult spawning run at the subbasin scale. I also note that these
sets of variables are present in the models that were identified
as best using BIC values. However, in isolation in models
5, 5b, 13, 14, and 14b, the relationships were not strongly
meaningful. This could reflect the lack of interaction between
these variables and the variables present in the other models.
Table 4.6. Bayesian Information Criterion (BIC) results for
adult coho salmon models from 2001 and 2002. BIC allows for
a comparison of model effectiveness by capturing the residual
variance and a penalty term. The model with the lowest BIC value
is considered to be the best fitting model. Models 5b and 14b
were created by altering the set of variables in models 5 and 14
respectively.
Year
Model Number
2001
2002
1
2
3
4
5
5b
6
7
8
9
10
11
12
13
14
14b
15
16
801.80
914.67
811.94
716.11
616.31
603.11
694.49
726.73
649.65
703.84
677.20
734.42
746.98
611.47
656.64
644.08
736.47
805.23
853.47
971.09
877.45
771.23
686.81
637.88
720.09
785.10
719.13
757.15
700.06
700.70
793.54
694.31
689.22
674.63
796.90
828.68
Flitcroft
Chapter 4
66
Table 4.7. Important combinations of variables from Bayesian hierarchical models of the mean of adult coho salmon at a subbasin scale.
Model
Number Parameters
2001
2002
0.31 (0.02 - 0.62)
0.44 (0.12 -0.79)
6
intercept for mean annual precipitation (basin scale) and percent large trees in
riparian buffer zone (subbasin scale)
7
intercept for mean annual precipitation (basin scale) and percent sedimentary
geology (subbasin scale)
0.58 (0.18 - 1.02)
8
intercept for mean annual precipitation (basin scale) and percent sedimentary
geology (subbasin scale)
0.48 (0.02 - 0.92)
13
intercept for mean annual precipitation (basin scale) and percent large trees in
riparian buffer zone (subbasin scale)
1.91 (0.21 - 4.47)
16
intercept for mean annual precipitation (basin scale) and percent sedimentary
geology (subbasin scale)
16
coefficient for number of tribs (basin scale) and mean annual precipitation
(subbasin scale)
Juveniles
A comparison of BIC values from initial modeling efforts
for juvenile coho salmon identified model 9 as the best (Table
4.8). In an attempt to improve this model, parameters in
the prior probability for the subbasin scale were successively
dropped and replaced with non-informative priors. In model
9b, percent subbasin in volcanic geology was dropped and
in model 9c percent subbasin in sedimentary geology was
dropped. Subbasin-scale variables were chosen because of
the simplicity of the original model, and because there was
no clear indication through the assessment of the parameter
coefficients that the specific subbasin variables contributed to
the explanatory power of the model. The replacement of the
subbasin scale variables with non-informative priors reduced
the number of estimated parameters in the model, and model
9c resulted in the lowest BIC value.
As with the adult models, I looked for meaningful parameters
that did not include 0 in their posterior interval. In none of
the juvenile models were subbasin level parameters identified
(meaning, all 95% posterior intervals for subbasin variables
contained zero). However, the presence of the subbasin level,
and therefore a hierarchical approach, was still important in
the organization of the information as reflected by variation
of the intercept parameter between subbasins. This variation
is visible in “caterpillar plots” that display the estimated values
and the 95% posterior intervals for all variables in a model
(Figure 4.4 from juvenile model #9c). Therefore, including
the subbasin scale using non-informative priors is appropriate
until a more specific relationship and variable is identified.
0.37 (0.09 - 0.77)
0.56 (0.24 - 0.88)
14.67 (0.52 - 28.99)
Table 4.8. Bayesian Information
Criterion (BIC) results for juvenile
coho salmon models from 2002. BIC
allows for a comparison of model
effectiveness by capturing the residual
variance and a penalty term. The model
with the lowest BIC value is considered
to be the best fitting model. Models
9b and 9c were created by altering
variables in model 9.
Model
Number
BIC
Results
1
2
3
4
5
6
7
8
9
9b
9c
10
3036.45
2635.68
2679.59
2802.46
2790.70
2814.73
2639.28
2671.80
2616.09
2609.42
2581.41
2669.43
Discussion
The juvenile and adult life history stages of coho salmon
have different habitat needs and levels of mobility that may
be conducive to separate analyses. Also, because of life history
differences, an exploration of different spatial scales for each
stage may be a biologically meaningful approach. Hierarchical
Bayesian modeling provided a method through which to
explore relationships between variables at different spatial
scales. This approach affirmed the usefulness of subbasin and
basin scales for modeling adult coho salmon, and sites and
subbasin scales for modeling juvenile coho salmon.
The first hypothesis explored in this paper focused on
adult coho salmon. I hypothesized that the spatial levels of
organization relevant for assessments of the abundance of
adult coho salmon spawners were the subbasin and basin
scales. Studies documenting the importance of subbasin
scale clumping of spawning adults (Neville et al. 2006) and
Flitcroft
Chapter 4
67
Figure 4.4. The caterpillar plot for juvenile
coho salmon model #9c displays the posterior
estimate and 95% interval for each parameter
in the model. These intervals may be interpreted
by examining their overlap and whether they
include zero. Intervals that include zero are
not considered to significantly contribute to
the model result. Intervals that do not overlap
display the importance of modeling including
the hierarchical scale of division. This allows
variability to be different with different
parameters rather than needing to be shared
throughout the dataset.
larger scale metapopulation level patterns analogous to the
basin scale (Cooper and Mangel 1999) provide context for
interpreting the results of this study.
The hierarchical model results lend support to the idea that
multiple levels of organization are reflected in the abundance
of adult coho salmon. Evidence for this may be found in the
recurring importance of two large-scale variables (percent large
trees in 100-m riparian buffer and mean annual precipitation)
in describing adult spawner abundance in several models over
two years (Table 4.7) and their presence in the model with the
lowest BIC (Table 4.4) in both 2001 and 2002. Riparian trees
at a subbasin scale were identified as important. The large scale
of a subbasin does not allow for an assessment of the spatial
relationship between spawning habitats, instream wood,
and source trees. However, riparian trees have been tied to
in-channel wood (Beechie and Sibley 1997), which is known
to accumulate gravels (Bilby and Ward 1989) necessary for
spawning (Hall and Baker 1982) and the formation of pool
habitats (Montgomery et al. 1995). Precipitation has long been
recognized as an important physical cue for adults to migrate
upstream to spawn (Allen 1959). Further, mean precipitation
in a basin may also be a surrogate value describing local climate
which will effect vegetation, and ultimately, the number and
size of riparian trees. Therefore, the significance of riparian
trees and precipitation variables in models of adult spawning
abundance are consistent with the biological needs of adult
coho salmon.
Developing and selecting a model with a low BIC is
useful, but it is also important to consider the context for this
model that is evident from assessments of all other models.
As previously mentioned, mean annual precipitation (basin
scale) and percent of trees in the riparian buffer (subbasin
scale) appeared as important descriptors of the mean of adult
spawners. Other variables that appeared to be important in
other models include percent sedimentary geology (subbasin
scale) paired with mean annual precipitation (basin scale) and
the number of tributaries (basin scale) also paired with mean
annual precipitation, this time at the subbasin scale (Table
4.7). The variable mean annual precipitation was almost
always the basin scale variable in the set of important variables.
This consistency at the basin scale indicates greater stability of
adult abundance at the basin scale with more variation at the
subbasin scale.
The Bayesian hierarchical analysis used in this research
confirms that the scales of subbasin and basin work together
in describing conditions conducive to adult coho occupancy.
Therefore, topology of dispersal by adults may best be described
by a combination of large and moderate spatial scales that
are analogous to population level dispersal. This conclusion
provides a framework of multiple large spatial scales for the
exploration of questions regarding persistence and habitat
use by adult coho salmon. Large spatial scales of adult coho
salmon may well be connected with large scale environmental
processes, such as disturbance, that effect survival of
Flitcroft
Chapter 4
individuals, and ultimately the entire population of salmon.
Disturbance will make some subbasins more or less hospitable
to populations of salmon over time. The adult behavior that
facilitates the consistent occupancy of available habitats is the
propensity of a small portion of the adult population to stray
(Labelle 1992, Nickelson 1998). Straying generally describes
the behavior of adults that do not return to their natal streams
to spawn. Adults that stray may return to a different part of
their natal basin, or to an entirely different stream system
to spawn, thereby allowing for the recolonization of new or
recovering habitats and the movement of genetic material
among subpopulations of fish. Future analysis that directly
explores large scale environmental conditions with behavioral
diversity of adult coho salmon may be useful in understanding
population level persistence of this species.
A foundation for further research of ecological processes
and spatial scale was set in this project by identifying spatial
scales that are useful for exploring the adult life history stage
of coho salmon. The analysis of adult distribution included
only two years of data and future work that includes more
years that span a variety of environmental conditions, such as
major flooding, drought, or large scale disturbances would be
worthwhile. Such temporal analysis would strengthen studies
over large regions and might include changes in basin-level
productivity in response to large-scale disturbance events.
Another hierarchical analysis of adults would benefit from the
incorporation of a site-specific scale because fine-scale habitat
variation has been shown to be important in the selection of
spawning sites (Neville et al. 2006).
The second hypothesis explored in this paper identified
site and subbasin scales for assessments of densities of
juvenile coho salmon. Juvenile coho must find all the habitats
necessary for them to survive for a year in freshwater before
migrating to the ocean as smolts. Habitat needs for juvenile
coho salmon change seasonally (Groot and Margolis 1991,
Quinn and Peterson 1996). Mainstem pools are important
in the summer, while off-channel and slow water habitats are
significant in the winter (Nickelson et al. 1992). Therefore,
local habitat conditions appear to be ecologically significant
for juvenile fish and this was confirmed by the selected
Bayesian model. The model with the lowest BIC incorporated
local variation without specifying relationships to higher
levels of spatial organization. The three variables in this model
were percent sand, stream order, and the network distance to
spawning habitat. The amount of fines, including sand, in a
pool is important for juvenile survival. High percentages of
fines decrease salmonid survival and growth in the first year by
shifting the food web towards taxa that are not prey food for
juvenile salmon (Suttle et al. 2004). Stream order is a useful
georeferencing metric because it places the pool within the
context of the stream system. For example, in Oregon’s coast
range, a pool that is located in a first order stream is generally
68
small with areas of high gradient. By contrast, a pool in a third
order stream may be wider and have lower gradients than the
first order stream, and is also downstream of several other
smaller tributaries that may deliver food or other nutrients
to the pool through drift (Vannote et al. 1980). Typically,
juvenile coho will have higher densities in lower-order stream
reaches than in higher order stream reaches (Rosenfeld et al.
2000). The network distance to spawning habitat is one of
the juxtaposition variables that references the pool to another
necessary habitat. Flitcroft (2007b) identified juxtaposition
variables as possibly important when exploring differences in
juvenile densities among subbasins over time, with subbasins
with large distances between seasonal habitats having lower
densities compared to basins with short distances. It is
interesting to note that the site-level variables in the juvenile
model with the lowest BIC included both substrate and
network juxtaposition metrics. The importance of these
variables is mirrored by the results described in Flitcroft
(2007b) in which juvenile coho salmon were best modeled
with a combination of instream and network variables. The
identification of similar variables in a Bayesian modeling
approach confirms the importance of both substrate and
network juxtaposition variables for describing juvenile density.
The combination of these metrics is ecologically informative
because they capture the local environmental conditions that
the juvenile fish are responding to.
While the inclusion of subbasin variables was not part of
the lowest BIC model of the density of juvenile coho salmon,
examining the caterpillar plot of the parameter estimates from
this model indicates that variation present at the intercept of
the model varies by subbasin. It would be useful to continue
exploring variables that represent subbasin variation because
I have clearly not identified the most important elements of
variation at the subbasin scale. The juvenile analysis would also
be enhanced by the assessment of more years of data, which
would help address the responsiveness of juveniles to changes
in larger scale environmental conditions over time.
The juvenile modeling results support my assumption
that the juxtaposition of seasonal habitats and local habitat
conditions are reflected by the topology of juvenile occupancy.
Further exploration of multiple years of juvenile site and
subbasin variables need to be incorporated in additional
research to properly explore the interplay between ecological
processes and spatial scale that juvenile coho salmon are
responding to. What was surprising in this analysis was the
lack of clear identification of any subbasin scale variables in
determining juvenile density. Beechie et al. (1994) identified the
larger scale variable of land use as an important consideration
in habitat loss that reduced coho salmon smolt output in the
Skagit River. Land use was not a variable that was incorporated
into this study, but is clearly an important descriptor of large
scale environmental conditions for coho salmon. It is also
Flitcroft
Chapter 4
possible that the way that I structured the hierarchical models
in this project did not capture the synergism among scales that
is important for juvenile coho salmon. More work exploring
specific variables at larger spatial scales needs to be done to
better understand the interaction between spatial scales on the
abundance of juvenile coho.
Conclusion
A hierarchical modeling approach provided a framework to
explore the complexity of spatial scales for juvenile and adult
coho salmon. The adult and juvenile life history stages of coho
salmon have different levels of mobility and drives. These
differences helped define the spatial scales that were used to
analyze these two life history stages. Considering the context
of spatial scale and stream network connectivity allowed
ecologically meaningful interpretations of the results. It is
evident that the interplay of spatial scales is different for adult
than for juvenile coho salmon. Juvenile coho salmon appear to
respond to local variation within the frame of larger subbasin
variation, while adult coho salmon are connected to subbasin
and basin variation and variables.
In this study, different life history stages were analyzed at
different scales which were intended to reflect population level
persistence strategies and the difference in mobility and habitat
needs between life history stages. Management that seeks to
restore, protect or enhance coho salmon populations must
consider the implications of scale and life history stage. In order
to maintain the capacity of coho salmon as a whole to survive,
the ability of the different life history stages to remain present
and vital in the population must be maintained. Considering
the spatial scale and extent occupied by each life history stage
is a platform for determining the needs of the fish. Juveniles
need great diversity of habitats in close proximity to one
another, while adults need access to a variety of spawning sites
within and among basins. Therefore, management that does
not consider each life history stage individually, and together,
may not adequately meet the survival needs of populations of
coho salmon.
Pacific salmon are adapted to survive in the dynamic stream
environment of the Pacific Northwest (Reeves et al. 1995).
Ecological and physiological processes may impact salmon
at different spatial scales. Joint consideration of population
needs and ecological processes through multiple spatial scales
makes tenable the management question of how population
persistence will be achieved in a continuously changing
environment.
69
Literature Cited
Allen, G.H. 1959. Behavior of Chinook and silver salmon. Ecology
40(1): 108-113.
Armstrong, J.D. 2005. Spatial variation in population dynamics
of juvenile Atlantic salmon: Implications for conservation and
management. Journal of Fish Biology 67 (Suppl. B): 35-52.
Beechie, T.E., E. Beamer and L. Wasserman. 1994. Estimating coho
salmon rearing habitat and smolt production losses in a large river
basin, and implications for habitat restoration. North American
Journal of Fisheries Management 14: 797-811.
Beechie, T.J. and T. H. Sibley. 1997. Relationships between channel
characteristics, woody debris, and fish habitat in northwestern
Washington streams. Transactions of the American Fisheries
Society 126: 217-229.
Bell, E. 2001. Survival, growth and movement of juvenile coho
salmon (Oncorhynchus kisutch) over-wintering in alcoves,
backwaters and main channel pools in Prairie Creek, California.
Master’s thesis. Humboldt State University, Arcata, California.
Bilby, R.E. and J.W. Ward. 1989. Changes in characteristics and
function of woody debris with increasing size of streams in western
Washington. Transactions of the American Fisheries Society 118:
368-378.
Bolton, S.M, J. Moss, J. Southard, G. Williams, C. DeBlois, N.
Evans and P. Wagner. 2002. Juvenile coho movement study.
Research Report Research Project T1803 Task 23 Fish Culvert
Passage. Washington State Transportation Commission, Seattle,
Washington.
Bradford, M.J., G.C. Taylor and J.A. Allan. 1997. Empirical review
of coho salmon smolt abundance and the prediction of smolt
production at the regional level. Transactions of the American
Fisheries Society 126: 49-64.
Bramblett, R.G., M.D. Bryant, B.E. Wright and R.G. White. 2002.
Seasonal use of small tributary and main-stem habitats by juvenile
steelhead, coho salmon, and Dolly Varden in a Southeastern
Alaska drainage basin. Transactions of the American Fisheries
Society 131: 498-506.
Browne, W.J. and D. Draper. 2006. A comparison of Bayesian and
likelihood-based methods for fitting multilevel models. Bayesian
Analysis 1(3): 473-514.
Carlin, B.P., J.S. Clark and A.E. Gelfand. 2006. Chapter 1. Elements
of hierarchical Bayesian inference. Pages 3-24 in J.S. Clarke and
A. E. Gelfand, Hierarchical Modelling for the Environmental
Sciences. Oxford University Press, New York.
Cooper, A.B. and M. Mangel. 1999. The dangers of ignoring
metapopulation structure for the conservation of salmonids.
Fisheries Bulletin 97: 213-226.
Daly, C., G.H. Taylor, W.P. Gibson, T.W. Parzybok, G.L. Johnson
and P.A. Pasteris. 1999. High-quality spatial climate data sets
for the United States and beyond. Transactions of the American
Society of Agricultural Engineers 43(6): 1957-1962.
Flitcroft
Chapter 4
70
Fausch, K.D., C.E. Torgersen, C.V. Baxter and H.W. Li. 2002.
Landscapes to Riverscapes: Bridging the gap between research and
conservation of stream fishes. Bioscience. 52(6) : 483-498.
Lee, D.C. and B.E. Rieman. 1997. Population viability assessment
of salmonids by using probabilistic networks. North American
Journal of Fisheries Management 17: 1144-1157.
Flitcroft, R.L. 2007a. Chapter 2: this volume. Oregon State
University, Corvallis Oregon.
Li, W.H. and twelve coauthors. 1995. Safe havens: Refuges and
evolutionarily significant units. Pages 371-380 in J.L. Nielsen,
editor. Evolution and the aquatic ecosystems: defining unique
units in population conservation. American Fisheries Society,
Symposium 17, Bethesda, Maryland.
Flitcroft, R.L. 2007b. Chapter 3: this volume. Oregon State
University, Corvallis Oregon.
Gelfand, A.E., A. Latimer, S. Wu and J.A. Silander, Jr. 2006. Building
statistical models to analyze species distributions. Pages 77-97 in
J.S. Clarke and A.E. Gelfand, editors. Hierarchical modeling for
the environmental sciences: statistical methods and applications.
Oxford University Press, New York.
Gelman, A., J.B. Carlin, H.S. Stern and D.B. Rubin. 2004. Bayesian
data analysis, 2nd Edition. Chapman and hall/CRC Press
Company, New York.
Gotway, C.A. and L.J. Young. 2002. Combining incompatible
spatial data. Journal of the American Statistical Association
97(458): 632-648.
Groot, C. and L. Margolis editors. 1991. Pacific Salmon Life
Histories. University of British Columbia Press,.Vancouver, B.C.,
Canada.
Hall, J.D and C.O. Baker 1982. Influence of forest and rangeland
management on anadromous fish habitat in western North
America: rehabilitating and enhancing stream habitat: Review
and evaluation. USDA Forest Service, Pacific Northwest Research
Station, Portland, Oregon. General Technical Report PNWGTR-138.
Harley, S.J. and R.A. Myers. 2001. Hierarchical Bayesian models
of length-specific catchability of research trawl surveys. Canadian
Journal of Fisheries and Aquatic Sciences 58: 1569-1584.
Jacobs, S., J. Firman, G. Susac, D. Steward and J. Weybright. 2002.
Status of Oregon coastal stocks of anadromous salmonids, 20002001 and 2001-2002. Oregon Department of Fish and Wildlife,
Report OPSW-ODFW-2002-3, Portland, Oregon.
Lichatowich, J. 1999. Salmon without rivers a history of the Pacific
salmon crisis. Island Press, Coveto, California.
Lowe, W.H., G.E. Likens and M.E. Power 2006. Linking scales in
stream ecology. BioScience 56 (7): 591-597.
Marcot, B.G., R.S. Holthausen, M.G. Raphael, M. Rowland and
M. Wisdom. 2001. Using Bayesian belief networks to evaluate
fish and wildlife population viability under land management
alternatives from an environmental impact statement. Forest
Ecology and Management 153(1-2): 29-42.
Matthews, W.J. and E. Marsh-Matthews. 2003. Effects of drought
on fish across axes of space, time and ecological complexity.
Freshwater Biology 48: 1232-1253.
Miller, D.J. 2003. Programs for DEM analysis. In landscape
dynamics and forest management. General Technical Report
RM-GTR-101CD. USDA Forest Service, Rocky Mountain
Research Station, Fort Collins, Colorado.
Monlean, V.J., A.I. Gitelman and A.N. Gray. 2002. Multi-scale
relationships between coarse woody debris and presence/absence
of Western Hemlock in the Oregon Coast Range. Lecture Notes
in Statistics 167: 311-318.
Montgomery, D.R., M.M. Buffington, R.D. Smith, K.M. Schmidt
and G. Pess. 1995. Pool spacing in forest channels. Water
Resources Research 31: 1097-1105.
Moore, K.M.S., K.K. Jones and J.M. Dambacher. 1997. Methods
for stream habitat surveys. Oregon Department of Fish and
Wildlife, Information Report 97-4, Portland, Oregon.
Jacobs, S.E. and R.E. Nickelson. 1989. Use of stratified random
sampling to estimate the abundance of Oregon coastal coho
salmon. Oregon Department of Fish and Wildlife, Report
F-145-R-09, Portland, Oregon.
Murray, C.B. and M.L. Rosenau. 1989. Rearing of juvenile Chinook
salmon in nonnatal tributaries of the lower Fraser River, British
Columbia. Transactions of the American Fisheries Society 118:
284-289.
Kahler, T.H., P. Roni and T.P. Quinn. 2001. Summer movement
and growth of juvenile anadromous salmonids in small western
Washington streams. Canadian Journal of Fisheries and Aquatic
Sciences 58: 1947-1956.
Myers, R.A., A. Mertz and J. Bridson. 1997. Spatial scales of
interannual recruitment variations of marine, anadromous,
and freshwater fish. Canadian Journal of Fisheries and Aquatic
Sciences 54: 1400-1407.
Labelle, M. 1992. Straying patterns of coho salmon (Oncorhynchus
kisutch) stocks from southeast Vancouver Island, British Columbia.
Canadian Journal of Fisheries and Aquatic Sciences 49: 18431855.
Neville, H.M., D.J. Isaak, J.B. Dunham, R.F. Thurow and B.E.
Rieman. 2006. Fine-scale natal homing and localized movement
as shaped by sex and spawning habitat in Chinook salmon: insights
from spatial autocorrelation analysis of individual genotypes.
Molecular Ecology 15: 4589-4602.
Lee, D.C. 2000. Chapter 9 Assessing land-use impacts on bull
trout using Bayesian belief networks. Pages 127-147 in Ferson
and Burzman. Quantitative methods for conservation biology.
Springer, New York.
Nickelson, T.E. 1998. A habitat-based assessment of coho salmon
production potential and spawner escapement needs for Oregon
coastal streams. Oregon Department of Fish and Wildlife
Information Report No. 98-4, Corvallis, Oregon.
Flitcroft
Chapter 4
Nickelson, T.E., J.D. Rodgers, S.L. Johnson and M.F. Solazzi.
1992. Seasonal changes in habitat use by juvenile coho salmon
Oncorhynchus kisutch in Oregon coastal streams. Canadian
Journal of Fisheries and Aquatic Sciences 49(4): 783-789.
Nielsen, J. 1992. Microhabitat-specific foraging behavior, diet, and
growth of juvenile coho salmon. Transactions of the American
Fisheries Society 121: 617-634.
Ohmann, J.L. and Gregory, M.J. 2002. Predictive mapping of
forest composition and structure with direct gradient analysis and
nearest-neighbor imputation in coastal Oregon, USA. Canadian
Journal of Forest Research 32: 725-741.
Pacific Fisheries Resource Conservation Council. 1999. Biological
stock structure: The staggering diversity in salmon population.
Salmon Stocks Background Paper No. 1999/1b.
Poff, N.L. and A.D. Huryn. 1998. Multi-scale determinants of
secondary production in Atlantic salmon (Salmo salar) streams.
Canadian Journal of Fisheries and Aquatic Sciences 55 (Suppl.
1): 201-217.
Quinn, T.P. 2005. The behavior and ecology of Pacific salmon and
trout. University of Washington Press, Seattle, Washington.
Quinn, T.P. and N.P. Peterson. 1996. The influence of habitat
complexity and fish size on over-winter survival and growth of
individually marked juvenile coho salmon (Oncorhynchus kisutch)
in Big Beef Creek, Washington. Canadian Journal of Fisheries and
Aquatic Sciences 53: 1555-1564.
Ramsey, F.L. and D.W. Schafer. 2002. The statistical sleuth: A course
in methods of data analysis. Second Edition. Duxbury Thomson
Learning, United States.
Redmond, K. and G. Taylor. 1997. Chapter 2 climate of the coastal
temperate rain forest. Pages 25-42 in Schoonmaker, P.K., B. von
Hagen and E.C. Wolf editors. The Rain forests of Home: Profile
of a North American Bioregion. Island Press, Washington, D.C.
71
Reeves, G.H., L.E. Benda, K.M. Burnett, P.A. Bisson and J.R. Sedell.
1995. A disturbance-based ecosystem approach to maintaining
and restoring freshwater habitats of evolutionarily significant units
of anadromous salmonids in the Pacific Northwest. American
Fisheries Society Symposium 17: 334-349.
Rosenfeld, J., M. Porter and E. Parkinson. 2000. Habitat factors
affecting the abundance and distribution of juvenile cutthroat
trout (Oncorhynchus clarki) and coho salmon (Oncorhynchus
kisutch). Canadian Journal of Fisheries and Aquatic Sciences 57:
766-774.
Spiegelhalter, D.A. Thomas, N. Best and D. Lunn. 2003. WinBUGS
User Manual Version 1.4. MRC Biostatistics Unit, Institute of
Public Health. Available: www.mrc-bsu.cam.ac.uk/bugs. (January
2007).
Suttle, K.B., M.E. Power, J.M. Levine and C. McNeely. 2004. How
fine sediment in riverbeds impairs growth and survival of juvenile
salmonids. Ecological Applications 14(4): 969-974.
Torgersen, C.E., R.E. Gresswell and D.S. Bateman. 2004. Pattern
detection in stream networks: Quantifying spatial variability in
fish distribution. GIS/Spatial Analyses in Fishery and Aquatic
Sciences pp. 405-420.
VerHoef, J.M. and K.J. Frost. 2003. A Bayesian hierarchical model
for monitoring harbor seal changes in Prince William Sound,
Alaska. Environmental and Ecological Statistics 10: 201-219.
Wiens, J.A. 1989. Spatial scaling in ecology. Functional Ecology 3:
385-397.
Wyatt, R.J. 2002. Estimating riverine fish population size from
single- and multiple-pass removal sampling using a hierarchical
model. Canadian Journal of Fisheries and Aquatic Sciences 59:
695-706.
Rebecca L. Flitcroft. 2007. Regions to streams: Spatial and temporal Variation in stream occupancy patterns of coho salmon (Oncorhynchus
kisutch) on the Oregon Coast. Ph.D. dissertation. Department of Fisheries and Wildlife, Oregon State University, Corvallis, OR.
CHAPTER 5
How Important are Network Considerations Anyway?
Rebecca L. Flitcroft
Department of Fisheries and Wildlife
Oregon State University
Corvallis, Oregon
Summary and Conclusions
Stream ecology is a relatively new science, so understandably,
the theoretical underpinnings continue to evolve. The research
described in this dissertation started as an investigation of the
implications of stream network structure on the abundance and
distribution of juvenile coho salmon (Oncorhynchus kisutch)
and evolved into the development of variables and theory
that include stream network relationships. The dissertation is
primarily focused on three key goals: the development and
application of a theoretical base for analyzing aquatic obligate
species that incorporates the underlying structure of the
stream network; exploring the importance of stream network
variables using multiple spatial scales in an assessment of
juvenile coho salmon density; incorporating multiple spatial
scales into hierarchical Bayesian models for the adult and
juvenile life history stage of coho salmon. The results of this
research point to further work, and also identify new ways of
interpreting and exploring patterns in the river landscape.
Chapter 2 begins with a discussion of the place of stream
networks in ecological theory and the current lack of a
stream network base from which to interpret and expand on
theory describing the movement of aquatic obligate species.
I propose and develop a dynamic network topology as a
framework for study design, analysis, and interpretation. The
dynamic network topology includes the idea that habitats
change over time in response to disturbance events or stream
flows and native aquatic species are adapted to these habitat
reconfigurations. The concept of dynamic network topology
moves beyond static predictions of aquatic communities
(Vannote et al. 1980) and integrates the consideration of
hierarchical processes (Frissell et al. 1986) into a dynamic
stream network (Benda et al. 1998, Poff et al. 1997).
In Chapter 3, I explore the importance of a stream network
context with an analysis of ecological processes affecting
juvenile coho salmon at three spatial scales (site, patch, and
subbasin). Multiple spatial scales have been identified as
important when exploring distributions (Schlosser 1995) and
productivity (Poff and Huryn 1998) of fishes. In the site scale
analysis, a combination of both network and instream metrics
are found to be important in understanding the patterns of
distribution of juvenile coho salmon. Further, the network
juxtaposition variables that describe the proximity between
habitats that are important seasonally for juvenile coho salmon
appear to differentiate between subbasins that have increasing
or decreasing juvenile densities between years. The proximity
between seasonal habitats has been identified as important for
Chinook (O. tshawytscha) (Isaak and Thurow 2006) and other
fishes (Jones et al. 2003). At the scale of a patch of juvenile
coho salmon, it appears that larger subbasins may have a
nested spatial structure in the distribution of juvenile coho
salmon. This nested structure could be associated with larger
quantities of habitats that might be found in larger stream
networks or could reflect an overlapping distribution of smaller
patches of juveniles. The subbasin scale analysis pointed to the
importance of the adult spawning run size in connection to
the number of stream km occupied by juvenile coho. In years
of low adult run size, the stream km occupied by juveniles
was lower than in years of higher run sizes. It is possible that
adults that spawn lower in the stream system in years of low
abundance are not occupying spawning habitats in the closest
proximity to the seasonal habitats necessary for their progeny.
This introduces some interesting questions about how adults
distribute themselves at spawning and what effect that may
have on the survival of their young.
Bayesian hierarchical methods are used in Chapter 4 to
structure models that are meant to represent two spatial scales
in analysis of the abundance of adult and juvenile coho salmon.
Multiple spatial scales allow for the interaction between
ecological processes that may occur in nature. Subbasin and
basin scales are hierarchically organized to explore adult coho
salmon abundance. These large spatial scales were chosen for
adults based on the mobility of adults, and because adults
define the distribution of coho salmon through homing and
straying behavior during the spawning season. The hierarchical
model that best fitted the adult data identified the variables:
percent of large trees in the riparian buffer at the subbasin
scale modeled with mean annual precipitation at the basin
scale. These variables are ecologically important for adult coho
salmon and are associated with processes of habitat creation
Flitcroft
Chapter 5
(Bilby and Ward 1989, Benda et al. 2004). Sites and subbasins
are used for juvenile coho salmon hierarchical models. These
smaller spatial scales are used for juveniles to reflect their need
to find seasonal habitats for a year of freshwater occupancy
in their natal stream network. Juveniles may move around
within their natal stream system (Kahler et al. 2001, Nielsen
1992), but they do not cross salt water in search of alternate
basins and habitats. This may mean that local scales best
reflect the movement and needs of juvenile coho salmon. The
hierarchical model that best fitted the juvenile data identified
a set of site level variables connected with non-informative
subbasin priors. The site scale variables were: percent sand,
stream order, and the network distance to spawning habitat.
These three variables capture local instream condition and the
network proximity of an important seasonal habitat.
The combination of a network perspective and multiple
spatial scales of analysis are a consistent theme in Chapters
2, 3, and 4 of this dissertation. This theme was intended to
create a framework that may better represent the impact of
stream processes on coho salmon abundance and distribution.
The results of the investigations in each chapter of this
dissertation support one another and confirm the usefulness
of a network perspective that incorporates multiple spatial
scales. The variables selected in the best fitting hierarchical
Bayesian model, in Chapter 4, for juvenile coho salmon
closely resemble the conclusions of the site scale investigation
in Chapter 3 that both instream and network variables
are important local descriptors of coho occupancy and
abundance. These interpretations confirm the premise of
Chapter 2 that considerations of aquatic species and ecological
processes within a stream network framework is a useful way
to approach analysis. Further, the use of multiple spatial scales
was useful in Chapter 3 for the interpretation of results and in
Chapter 4 as the model structure itself. The combination of a
stream network perspective at multiple scales resulted in novel
applications of statistical method and ecological interpretation.
Throughout this dissertation, the structure of the stream as
an interconnected network guided ecological interpretation,
statistical approaches, and study design. The contribution
of this work for ecological investigation is two-fold. First,
the focus on integrating a stream network framework into
considerations of aquatic life history stages at multiple spatial
scales is novel and proved useful. Second, using analytic
approaches that incorporate network structure is unusual and
may be useful for other investigations of aquatic species. The
ultimate goal for any ecological investigation is to contribute
to the body of work that describes the natural world in a
meaningful and thoughtful manner. It is in this vein that the
work of this dissertation has been undertaken.
73
Literature Cited
Benda, L., D. Miller, T. Dunne, J. Agee and G. Reeves. 1998.
Dynamic landscape systems, Chapter 12. Pages 261-288 in R.J.
Naiman and R.E. Bilby, editors. River ecology and management:
lessons from the Pacific Coastal Ecoregion. Springer-Verlag, New
York.
Benda, L., L. Poff, D. Miller, T. Dunne, G. Reeves, G. Pess and M.
Pollock. 2004. The network dynamics hypothesis: How channel
networks structure riverine habitats. BioScience 54(5): 413-427.
Bilby, R.E. and J.W. Ward. 1989. Changes in characteristics and
function of woody debris with increasing size of streams in western
Washington. Transactions of the American Fisheries Society 118:
368-378.
Frissell, C.A., W.J. Liss, C.E. Warren and M.D. Hurley. 1986. A
hierarchical framework for stream habitat classification: viewing
streams in a watershed context. Environmental Management
10(2): 199-214.
Isaak, D.J., and R.F. Thurow. 2006. Network-scale spatial and
temporal variation in Chinook salmon (Oncorhynchus tshawytscha)
redd distributions: Patterns inferred from spatially continuous
replicate surveys. Canadian journal of Fisheries and Aquatic
Sciences 63(2): 285-296.
Jones, M.L., J.K. Netto, J.D. Stockwell, and J.B. Mion. 2003.
Does the value of newly accessible spawning habitat for walleye
(Stizostedion vitreum) depend on its location relative to nursery
habitats? Canadian Journal of Fisheries and Aquatic Sciences 60:
1527-1538.
Kahler, T.H., P. Roni and T.P. Quinn. 2001. Summer movement
and growth of juvenile anadromous salmonids in small western
Washington streams. Canadian Journal of Fisheries and Aquatic
Sciences 58: 1947-1956.
Nielsen, J. 1992. Microhabitat-specific foraging behavior, diet, and
growth of juvenile coho salmon. Transactions of the American
Fisheries Society 121: 617-634.
Poff, N.L., J.D. Allan, M.B. Bain, J.R. Karr, K.L. Prestegaard, B.D.
Richter, R.E. Sparks and J.C. Stromberg. 1997. The natural
flow regime a paradigm for river conservation and restoration.
BioScience 47(11): 769-784.
Poff, N.L. and A.D. Huryn. 1998. Multi-scale determinants of
secondary production in Atlantic salmon (Salmo salar) streams.
Canadian Journal of Fisheries and Aquatic Sciences 55 (Suppl.
1): 201-217.
Schlosser, I.J. 1995. Critical landscape attributes that influence fish
population dynamics in headwater streams. Hydrobiologia 303:
71-81.
Vannote, R.L., G.W. Minshall, K.W. Cummins, J.R. Sedell and C.E.
Cushing. 1980. The river continuum concept. Canadian Journal
of Fisheries and Aquatic Science 37: 130-137.
Bibliography
Allen, G.H. 1959. Behavior of Chinook and silver salmon. Ecology
40(1): 108-113.
Argent, D.G., J.A. Bishop, J.R. Stauffer, Jr., R.F. Carline and W.L.
Myers. 2003. Predicting freshwater fish distributions using
landscape-level variables. Fisheries Research 60: 17-32.
Armstrong, J.D. 2005. Spatial variation in population dynamics
of juvenile Atlantic salmon: Implications for conservation and
management. Journal of Fish Biology 67 (Suppl. B): 35-52.
Beechie, T.J. and T.H. Sibley. 1997. Relationships between channel
characteristics, woody debris, and fish habitat in northwestern
Washington streams. Transactions of the American Fisheries
Society 126: 217-229.
Bell, E. 2001. Survival, growth and movement of juvenile coho
salmon (Oncorhynchus kisutch) over-wintering in alcoves,
backwaters and main channel pools in Prairie Creek, California.
Master’s thesis. Humboldt State University, Arcata, California.
Bell, E, W.G. Duffy and T.D. Roelofs. 2001. Fidelity and survival of
juvenile coho salmon in response to a flood. Transactions of the
American Fisheries Society 130: 450-458.
Benda, L. and T. Dunne. 1997. Stochastic forcing of sediment
supply to channel networks from landsliding and debris flow.
Water Resources Research 33(12): 2849-2863.
Benda, L., D. Miller, T. Dunne, J. Agee and G. Reeves. 1998.
Dynamic landscape systems, Chapter 12. Pages 261-288 in R.J.
Naiman and R.E. Bilby, editors. River ecology and management:
Lessons from the Pacific Coastal Ecoregion. Springer-Verlag, New
York.
Benda, L., L. Poff, D. Miller, T. Dunne, G. Reeves, G. Pess and M.
Pollock. 2004. The network dynamics hypothesis: how channel
networks structure riverine habitats. BioScience 54(5): 413-427.
Bilby, R.E. and J.W. Ward. 1989. Changes in characteristics and
function of woody debris with increasing size of streams in western
Washington. Transactions of the American Fisheries Society 118:
368-378.
Bisson, P.A., K. Sullivan, and J.L. Nielsen. 1988. Channel hydraulics,
habitat use, and body form of juvenile coho salmon, steelhead, and
cutthroat trout in streams. Transactions of the American Fisheries
Society 117(3): 262-273.
Bolton, S.M, J. Moss, J. Southard, G. Williams, C. DeBlois, N.
Evans and P. Wagner. 2003. Juvenile coho movement study.
Research Report Research Project T1803 Task 23 Fish Culvert
Passage. Washington State Transportation Commission, Seattle,
Washington.
Bradford, M. J., Taylor G. C. and J. A. Allan. 1997. Empirical review
of coho salmon smolt abundance and the prediction of smolt
production at the regional level. Transactions of the American
Fisheries Society 126: 49-64.
Bramblett, R.G., M.D. Bryant, B.E. Wright and R.G. White. 2002.
Seasonal use of small tributary and main-stem habitats by juvenile
steelhead, coho salmon, and Dolly Varden in a Southeastern
Alaska drainage basin. Transactions of the American Fisheries
Society 131: 498-506.
Browne, W. J. and D. Draper. 2006. A comparison of Bayesian and
likelihood-based methods for fitting multilevel models. Bayesian
Analysis 1(3): 473-514.
Burnett, K.M. 2001. Relationships among juvenile anadromous
salmonids, their freshwater habitat, and landscape characteristics
over multiple years and spatial scales in the Elk River, Oregon.
Doctoral dissertation. Oregon State University, Corvallis, Oregon.
Burnett, K.M., G.H. Reeves, D.J. Miller, S. Clarke, K. VanceBorland and K. Christiansen. 2007. Distribution of salmon-habitat
potential relative to landscape characteristics and implications for
conservation. Ecological Applications 17(1): 66-80.
Carlin, B.P., J.S. Clark and A.E. Gelfand. 2006. Chapter 1. Elements
of hierarchical Bayesian inference. Pages 3-24 in J.S. Clarke and
A.E. Gelfand, Hierarchical Modelling for the Environmental
Sciences. Oxford University Press, New York.
Chapman, D.W. 1962. Food and space as regulators of salmonid
populations in streams. The American Naturalist 100(913): 345357.
Church, M. 2002. Geomorphic thresholds in Riverine landscapes.
Freshwater Biology 47: 541-557.
Clarke, S.E., K.M. Burnett and D.J. Miller. In review. Modeling
streams and hydrogeomorphic attributes in Oregon from digital
and field data. Journal of American Water Resources Research.
Collen, P. and R.J. Gibson. 2001. The general ecology of beavers
(Castor spp.), as related to their influence on stream ecosystems
and riparian habitats, and the subsequent effects on fish – A
review. Reviews in Fish Biology and Fisheries 10: 439-461.
Cooney, C.X. and S.E. Jacobs. 1995. Oregon coastal salmon
spawning surveys, 1993. Oregon Department of Fish and Wildlife,
Fish Division, Portland, Oregon.
Cooper, A.B. and M. Mangel. 1999. The dangers of ignoring
metapopulation structure for the conservation of salmonids.
Fisheries Bulletin 97: 213-226.
Cressie, N., J. Frey, B. Harch and M. Smith. 2006. Spatial prediction
on a river network. Journal of Agricultural, Biological and
Environmental Statistics 11(2): 127-150.
Daly, C., G.H. Taylor, W.P. Gibson, T.W. Parzybok, G.L. Johnson
and P.A. Pasteris. 1999. High-quality spatial climate data sets
for the United States and beyond. Transactions of the American
Society of Agricultural Engineers 43(6): 1957-1962.
Day, D.M., W.A. Bertrand, M.J. Wiley and R. Sauer. 1992.
Influence of stream location in a drainage network on the index
of biotic integrity. Transactions of the American Fisheries Society
121: 635-643.
Dunham, J.B, B.E. Rieman, and J.T. Peterson. 2002. Chapter 26:
Patch-based models to predict occurrence: lessons from salmonid
Flitcroft
Bibliography
fishes in streams. Pages 327-334. in J.M. Scott, P. Heglund,
M. Morrison, editors. Predicting species occurrences issues of
accuracy and scale. Island Press, Washington, D.C.
Ettema, C.H. and D.A. Wardle. 2002. Spatial soil ecology. TRENDS
in Ecology and Evolution 17(4): 177-173
Fagan, W.F. 2002. Connectivity, fragmentation, and extinction risk
in dendritic metapopulations. Ecology 83(23): 3243-3249.
Fausch, K.D., C.E. Torgersen, C.V. Baxter and H.W. Li. 2002.
Landscapes to riverscapes: Bridging the gap between research and
conservation of stream fishes. Bioscience 52(6): 483-498.
Fisher, S.G. 1997. Creativity, idea generation, and the functional
morphology of streams. Journal of the North American
Benthological Society 16: 305-318.
Flitcroft, R.L. 2007. Chapter 2: this volume. Oregon State
University, Corvallis Oregon.
Flitcroft, R.L. 2007. Chapter 3: this volume. Oregon State
University, Corvallis Oregon.
75
prediction of salmonid abundance in streams. Canadian Journal
of Aquatic Science 55 (Supplement 1): 181-190.
Grenouillet, G., D. Pont and C. Herisse. 2004. Within-basin fish
assemblage structure: the relative influence of habitat versus
stream spatial position on local species richness. Canadian Journal
of Fisheries and Aquatic Sciences 61: 93-102.
Gresswell, R.E., C.E. Torgersen and D.S. Bateman. 2006. A
spatially explicit approach for evaluating relationships among
coastal cutthroat trout, habitat, and disturbance in headwater
streams. Pages 457-471 in R.M. Hughes, L. Wang and P.W.
Seelbach, editors. Influences of landscapes on stream habitats and
biological assemblages. American Fisheries Society, Symposium
48, Bethesda, Maryland.
Groot, C. and L. Margolis, editors. 1991. Pacific Salmon Life
Histories. University of British Columbia Press, Vancouver, B.C.,
Canada.
Foster, S.C., C.H. Stein and K.K. Jones. 2001. A guide to
interpreting stream survey reports. Edited by P.A. Bowers. Oregon
Department of Fish and Wildlife, Information Reports 2001-06,
Portland, Oregon.
Hall, J.D. and C.O. Baker 1982. Influence of forest and rangeland
management on anadromous fish habitat in western North
America: rehabilitating and enhancing stream habitat: Review
and evaluation. USDA Forest Service, Pacific Northwest Research
Station: Portland, Oregon. General Technical Report PNWGTR-138.
Frissell, C.A., W.J. Liss, C.E. Warren and M.D. Hurley. 1986. A
hierarchical framework for stream habitat classification: viewing
streams in a watershed context. Environmental Management
10(2): 199-214.
Hankin, D.G. and G.H. Reeves. 1988. Estimating total fish
abundance and total habitat area in small streams based on visual
estimation methods. Canadian Journal of Fisheries and Aquatic
Sciences 45: 834-844.
Ganio, L.M., C.E. Torgersen and R.E. Gresswell. 2005. A
geostatistical approach for describing spatial pattern in stream
networks. Frontiers in Ecology and the Environment 3(3): 138144.
Hanski, I. and M. Gilpin. 1991. Metapopulation dynamics: A brief
history and conceptual domain. Biological Journal of the Linnean
Society 42: 3-16.
Gardner, B., P.J. Sullivan and A.J. Lembo, Jr. 2003. Predicting
stream temperatures: geostatistical model comparison using
alternative distance metrics. Canadian Journal of Fisheries and
Aquatic Sciences 60: 344-351.
Gelfand, A.E., A. Latimer, S. Wu and J.A. Silander, Jr. 2006. Building
statistical models to analyze species distributions. Pages 77-97 in
J.S. Clarke and A.E. Gelfand, editors. Hierarchical modeling for
the environmental sciences: statistical methods and applications.
Oxford University Press, New York.
Gelman, A., J.B. Carlin, H.S. Stern and D.B. Rubin. 2004. Bayesian
data analysis, 2nd Edition. Chapman and Hall/CRC Press
Company, New York.
Gerking, S.D. 1959. The restricted movement of fish populations.
Biological Review 34: 221-242.
Gleason, H.A. 1926. The individualistic concept of the plant
association. Torreya 53: 7-26.
Gleason, H.A. 1927. Further views of the succession concept.
Ecology 8: 299-326.
Gotway, C.A. and L.J. Young. 2002. Combining incompatible
spatial data. Journal of the American Statistical Association
97(458): 632-648.
Grant, J.W.A., S.O. Steingrimsson, E.R. Keeley and R.A. Cunjak.
1998. Implications of territory size for the measurement and
Harley, S.J. and R.A. Myers. 2001. Hierarchical Bayesian models
of length-specific catchability of research trawl surveys. Canadian
Journal of Fisheries and Aquatic Sciences 58: 1569-1584.
Heggenes, J., T.G. Northcote and A. Peter. 1991. Spatial stability of
cutthroat trout (Oncorhynchus clarki) in a small, coastal stream.
Canadian Journal of Fisheries and Aquatic Sciences 48: 757-762.
Horn, H.S. 1981. Chapter 11: Succession. Pages 253-271 in May,
R.M., editor. Theoretical Ecology: Principles and Applications.
Blackwell Scientific Publications, Boston.
Isaak, D.J. and R.F. Thurow. 2006. Network-scale spatial and
temporal variation in Chinook salmon (Oncorhynchus tshawytscha)
redd distributions: Patterns inferred from spatially continuous
replicate surveys. Canadian Journal of Fisheries and Aquatic
Sciences 63(2): 285-296.
Jacobs, S.E. and C.X. Cooney. 1995. Improvement of methods used
to estimate the spawning escapement of Oregon coastal natural
coho salmon. Oregon Department of Fish and Wildlife, Portland,
Oregon.
Jacobs, S., J. Firman, G. Susac, D. Steward and J. Weybright. 2002.
Status of Oregon coastal stocks of anadromous salmonids, 20002001 and 2001-2002. Oregon Department of Fish and Wildlife,
Report OPSW-ODFW-2002-3, Portland, Oregon.
Jacobs, S.E. and R.E. Nickelson. 1989. Use of stratified random
sampling to estimate the abundance of Oregon coastal coho
Flitcroft
Bibliography
salmon. Oregon Department of Fish and Wildlife, Report
F-145-R-09, Portland, Oregon.
Johnson, S.L. 1999. Fish habitat restoration in tenmile Creek lifecycle monitoring project summary report. Oregon Department of
Fish and Wildlife, Salem, Oregon.
Jones, K.K., J.M. Dambacher and R.L. Flitcroft. 2007. Effectiveness
and applicability of EMAP survey design in status review of Great
Basin redband trout. Extended Abstract to be published as part of
2007 Redband Workshop.
Jones, K.K. and K.M.S. Moore. 1999. Habitat assessment in coastal
basins in Oregon: implications for coho salmon production
and habitat restoration. Pages 329-340 in E.E. Knudsen, C.R.
Steward, D.D. McDonald, J.E. Williams and D.W. Riser, Editors.
Sustainable Fisheries Management. CRC Press, New York.
Jones, M.L., J.K. Netto, J.D. Stockwell, and J.B. Mion. 2003.
Does the value of newly accessible spawning habitat for walleye
(Stizostedion vitreum) depend on its location relative to nursery
habitats? Canadian Journal of Fisheries and Aquatic Sciences 60:
1527-1538.
Joy, M.K. and R.G. Death. 2004. Predictive modeling and spatial
mapping of freshwater fish and decapod assemblages using GIS
and neural networks. Freshwater Biology 49: 1036-1052.
Kahler, T.H., P. Roni and T.P. Quinn. 2001. Summer movement
and growth of juvenile anadromous salmonids in small western
Washington streams. Canadian Journal of Fisheries and Aquatic
Sciences 58: 1947-1956.
Kocik, J.F. and C.P. Ferreri 1998. Juvenile production variation in
salmonids: population dynamics, habitat, and the role of spatial
relationships. Canadian Journal of Fisheries and Aquatic Science
55(Suppl. 1): 191-200.
Kocovsky, P.M, and R.F. Carline. 2006. Influence of landscapescale factors in limiting brook trout populations in Pennsylvania
streams. Transactions of the American Fisheries Society 135;
76-88.
Labelle, M. 1992. Straying patterns of coho salmon (Oncorhynchus
kisutch) stocks from southeast Vancouver Island, British Columbia.
Canadian Journal of Fisheries and Aquatic Sciences 49: 18431855.
Landres, P.B., P. Morgan and F.J. Swanson. 1999. Overview of the
use of natural variability concepts in managing ecological systems.
Ecological Applications 9(4): 1179-1188.
Larsen, D.P., T.M. Kincaid, S.E. Jacobs and N.S. Urquhart. 2001.
Designs for evaluating local and regional scale trends. BioScience
51(12): 1069-1078.
Lee, D.C. 2000. Chapter 9. Assessing land-use impacts on bull
trout using Bayesian belief networks. Pages 127-147 in Ferson
and Burzman. Quantitative Methods for Conservation Biology.
Springer, New York.
Lee, D.C. and B.E. Rieman. 1997. Population viability assessment
of salmonids by using probabilistic networks. North American
Journal of Fisheries Management 17: 1144-1157.
Legendre, P. and M.J. Fortin. 1989. Spatial pattern and ecological
76
analysis. Vegetatio 80: 107-138.
Levin, S.A. 1992. The problem of pattern and scale in ecology: The
Robert H. MacArthur award lecture. Ecology 73(6): 1943-1967.
Li, W.H. and twelve coauthors. 1995. Safe havens: Refuges and
evolutionarily significant units. Pages 371-380 in J.L. Nielsen,
editor. Evolution and the aquatic ecosystems: Defining unique
units in population conservation. American Fisheries Society,
Symposium 17, Bethesda, Maryland.
Lichatowich, J. 1999. Salmon without rivers a history of the Pacific
salmon crisis. Island Press, Coveto, California.
Lichatowich, J., L. Mobrand, L. Lestelle and T. Vogel. 1995. An
approach to the diagnosis and treatment of depleted Pacific
salmon populations in freshwater ecosystems. Fisheries (Bethesda)
20(l): 10-18.
Lonzarich, D.G., M.R. Lonzarich and J.L. Warren, Jr. 2000. Effects
of riffle length on the short-term movement of fishes among
stream pools. Canadian Journal of Fisheries and Aquatic Sciences
57: 1508-1514.
Lowe, W.H., G.E. Likens and M.E. Power. 2006. Linking scales in
stream ecology. BioScience 56(7): 591-597.
Lunetta, R.S., B.L. Cosentino, D.R. Montgomery, E.M. Beamer
and T.J. Beechie. 1997. GIS-based evaluation of salmon habitat in
the Pacific Northwest. Photogrammetric Engineering & Remote
Sensing 63(10): 1219-1229.
Marcot, B.G., R.S. Holthausen, M.G. Raphael, M. Rowland and
M. Wisdom. 2001. Using Bayesian belief networks to evaluate
fish and wildlife population viability under land management
alternatives from an environmental impact statement. Forest
Ecology and Management 153(1-2): 29-42.
Matthews, W.J. and E. Marsh-Matthews. 2003. Effects of drought
on fish across axes of space, time and ecological complexity.
Freshwater Biology 48: 1232-1253.
McCune, B. 2004. Nonparametric multiplicative regression for
habitat modeling. Available: http://www.pcord.com/NPMRinto.
pdf. (January 2007).
McCune, B. and M.J. Mefford. 1999. PC-ORD. Multivariate
analysis of ecological data. Version 4.33 MjM Software, Gleneden
Beach, Oregon.
McCune, B. and M.J. Mefford. 2004. Hyperniche. Nonparametric
multiplicative habitat modeling. Version 1.05 MjM Software,
Gleneden Beach, Oregon.
Megalhaes, M.F., D.C. Batalha and M.J. Collares-Pereira. 2002.
Gradients in stream fish assemblages across a Mediterranean
landscape: contributions of environmental factors and spatial
structure. Freshwater Biology 47: 1015-1031.
Miller, D.J. 2003. Programs for DEM analysis. In landscape
dynamics and forest management. General Technical Report
RM-GTR-101CD. USDA Forest Service, Rocky Mountain
Research Station, Fort Collins, Colorado.
Miller, D.J., C. Luce and L. Benda. 2003. Time, space, and
episodicity of physical disturbance in streams. Forest Ecology and
Management 178: 121-140.
Flitcroft
Bibliography
Monlean, V.J., A.I. Gitelman and A.N. Gray. 2002. Multi-scale
relationships between coarse woody debris and presence/absence
of Western Hemlock in the Oregon Coast Range. Lecture Notes
in Statistics 167: 311-318.
Montgomery, D.R., E.M. Beamer, G.R. Pess and T.P. Quinn. 1999.
Channel type and salmonid spawning distribution and abundance.
Canadian Journal of Fisheries and Aquatic Sciences 56: 377-387.
Montgomery, D.R., M.M. Buffington, R.D. Smith, K.M. Schmidt
and G. Pess. 1995. Pool spacing in forest channels. Water
Resources Research 31: 1097-1105.
Moore, K.M.S., K.K. Jones and J.M. Dambacher. 1997. Methods
for stream habitat surveys. Oregon Department of Fish and
Wildlife, Information Report 97-4, Portland, Oregon.
Murray, C.B. and M.L. Rosenau. 1989. Rearing of juvenile Chinook
salmon in nonnatal tributaries of the lower Fraser River, British
Columbia. Transactions of the American Fisheries Society 118:
284-289.
Myers, R.A., A. Mertz and J. Bridson. 1997. Spatial scales of
interannual recruitment variations of marine, anadromous,
and freshwater fish. Canadian Journal of Fisheries and Aquatic
Sciences 54: 1400-1407.
Naiman, R.J., T.J. Beechie, L.E. Benda, D.R. Berg, P.A. Bisson,
L.H. MacDonald, M.D. O’Connor, P.L. Olson and E.A. Steel.
1992. Fundamental elements of ecologically healthy watersheds
in the Pacific Northwest coastal ecoregion. Pages 127-187 in R.J.
Naiman, editor. Watershed Management: Balancing Sustainability
and Environmental Change. Springer-Verlag, New York.
Nakamura, F., F.J. Swanson and S.M. Wondzell. 2000. Disturbance
regimes of stream and riparian systems – A disturbance-cascade
perspective. Hydrological Processes 14: 2849-2860.
Neville, H.M., D.J. Isaak, J.B. Dunham, R.F. Thurow and B.E.
Rieman. 2006. Fine-scale natal homing and localized movement as
shaped by sex and spawning habitat in Chinook salmon: Insights
from spatial autocorrelation analysis of individual genotypes.
Molecular Ecology 15: 4589-4602.
Nickelson, T.E. 1998. A habitat-based assessment of coho salmon
production potential and spawner escapement needs for Oregon
coastal streams. Oregon Department of Fish and Wildlife
Information Report No. 98-4, Corvallis, Oregon.
Nickelson, T.E., M.F. Solazzi, S.L. Johnson and J.D. Rodgers.
1992a. An approach to determining stream carrying capacity and
limiting habitat for coho salmon Oncorhynchus kisutch. Pages 251260 in L. Berg and P.W. Delaney, editors. Proceedings of the coho
workshop. Naniamo, B.C., Canada.
Nickelson, T.E., J.D. Rodgers, S.L. Johnson and M.F. Solazzi.
1992b. Seasonal changes in habitat use by juvenile coho salmon
Oncorhynchus kisutch in Oregon coastal streams. Canadian Journal
of Fisheries and Aquatic Sciences 49(4): 783-789.
Nielsen, J. 1992. Microhabitat-specific foraging behavior, diet, and
growth of juvenile coho salmon. Transactions of the American
Fisheries Society 121: 617-634.
Ohmann, J.L. and Gregory, M.J. 2002. Predictive mapping of
forest composition and structure with direct gradient analysis and
77
nearest-neighbor imputation in coastal Oregon, USA. Canadian
Journal of Forest Research 32: 725-741.
Oregon Department of Fish and Wildlife. 2005. Oregon list of
threatened and endangered fish and wildlife species. Available:
http://www.dfw.state.or.us/
threatened_endangered/t_e.html
(July 2007).
Ottaway, E.M. and D.R. Forrest. The influence of water velocity
on the downstream movement of alevins and fry of brown trout,
Salmo trutta L. Journal of Fish Biology 23(2): 221-227.
Pacific Fisheries Resource Conservation Council. 1998. Biological
stock structure: The staggering diversity in salmon population.
Salmon Stocks Background Paper No. 1999/1b.
Peters, D.P.C., J.R. Gosz, W.T. Pockman, E.E. Small, R.R.
Parmenter, S.L. Collins and E. Muldavin. 2006. Integrating
patch and boundary dynamics to understand and predict biotic
transitions at multiple scales. Landscape Ecology 21: 19-33.
Poff, N.L., J.D. Allan, M.B. Bain, J.R. Karr, K.L. Prestegaard, B.D.
Richter, R.E. Sparks and J.C. Stromberg. 1997. The natural
flow regime a paradigm for river conservation and restoration.
BioScience 47(11): 769-784.
Poff, N.L. and A.D. Huryn. 1998. Multi-scale determinants of
secondary production in Atlantic salmon (Salmo salar) streams.
Canadian Journal of Fisheries and Aquatic Sciences 55 (Suppl.
1): 201-217.
Pollock, M.M., G.R. Pess and T.J. Beechie. 2004. The importance of
beaver ponds to coho salmon production in Stillaquamish River
Basin, Washington, USA. North American Journal of Fisheries
Management 24: 749-760.
Quinn, T.P. 2005. The behavior and ecology of Pacific salmon and
trout. University of Washington Press, Seattle, Washington.
Quinn, T.P. and N.P. Peterson. 1996. The influence of habitat
complexity and fish size on over-winter survival and growth
of individually marked juvenile coho salmon (Oncorhynchus
kisutch) in Big Beef Creek, Washington. Canadian Journal of
Fisheries and Aquatic Sciences 53: 1555-1564.
Ramsey, F.L. and D.W. Schafer. 2002. The statistical sleuth: A course
in methods of data analysis. Second Edition. Duxbury Thomson
Learning, U.S.
Redmond, K. and G. Taylor. 1997. Chapter 2: Climate of the coastal
temperate rain forest. Pages 25-42 in Schoonmaker, P.K., B. von
Hagen, and E.C. Wolf, editors. The Rain Forests of Home: Profile
of a North American Bioregion. Island Press, Washington, D.C.
Reeves, G.H., L.E. Benda, K.M. Burnett, P.A. Bisson and J.R. Sedell.
1995. A disturbance-based ecosystem approach to maintaining
and restoring freshwater habitats of evolutionarily significant units
of anadromous salmonids in the Pacific Northwest. American
Fisheries Society Symposium 17: 334-349.
Reeves, G.H., K.M. Burnett and S.V. Gregory. 2002. Fish and
aquatic ecosystems of the Oregon Coast Range. in Hobbs, S.D.,
J.P. Hayes, R.L. Johnson, G.H. Reeves, T.A Spies, J.C. Tappeiner
II, and G.E. Wells, editors. Forest and stream management in the
Oregon Coast Range. Oregon State University Press, Corvallis,
Oregon.
Flitcroft
Bibliography
Remington, K. 1999. ArcScript: shortest network paths v1.1.
Available:
http://arcscripts.esri.com/details.asp?dbid=10895.
(January 2007).
Richards, C. and P. J. Cernera. 1989. Dispersal and abundance
of hatchery-reared and naturally spawned juvenile Chinook
salmon in an Idaho stream. North American Journal of Fisheries
Management 9: 345-351.
Rieman, B.E. and J.B. Dunham. 2000. Metapopulations and
salmonids: A synthesis of life history patterns and empirical
observations. Ecology of Freshwater Fish 9: 51-64.
Rosenfeld, J., M. Porter and E. Parkinson. 2000. Habitat factors
affecting the abundance and distribution of juvenile cutthroat
trout (Oncorhynchus clarki) and coho salmon (Oncorhynchus
kisutch). Canadian Journal of Fisheries and Aquatic Sciences 57:
766-774.
Rosgen, D.L. 1994. A classification of natural rivers. Catena 22:
169-199.
Schlosser, I.J. 1985. Flow regime, juvenile abundance, and the
assemblage structure of stream fishes. Ecology 66(5): 1484-1490.
Schlosser, I. J. 1991. Stream fish ecology: A landscape perspective.
BioScience 41(10): 704-712.
Schlosser, I.J. 1995. Critical landscape attributes that influence fish
population dynamics in headwater streams. Hydrobiologia 303:
71-81.
Scott, J.M., B. Csuti, J.D. Jacobi and J.E. Estes. 1987. Species
richness a geographic approach to protecting future biological
diversity. BioScience 37(11): 782-788.
Shapovalov, L. and A.C. Taft. 1954. The life histories of the steelhead
rainbow trout (Salmo gairdneri gairdneri) and silver salmon
(Oncorhynchus kisutch) with special reference to Wadel Creek,
California, and recommendations regarding their management.
Fish Bulletin of the California Department of Fish and Game
98:1-375.
Sleeper, J. 1993. Seasonal changes in distribution and abundance
of salmonids and habitat availability in a coastal Oregon basin.
Masters thesis. Oregon State University, Corvallis, Oregon.
Smith, T.A. and C.E. Kraft. 2005. Stream fish assemblages in relation
to landscape position and local habitat variables. Transactions of
the American Fisheries Society 134: 430-440.
Southwood, T.R.E. 1977. Habitat, the templet for ecological
strategies? The Journal of Animal Ecology 46(2): 336-365.
Spalding, S., N.P. Peterson and T.P. Quinn. 1995. Summer
distribution, survival, and growth of juvenile coho salmon
under varying experimental conditions of brushy instream cover.
Transactions of the American Fisheries Society 124: 124-130.
Spiegelhalter, D.A. Thomas, N. Best and D. Lunn. 2003. WinBUGS
User Manual Version 1.4. MRC Biostatistics Unit, Institute of
Public Health. Available: www.mrc-bsu.cam.ac.uk/bugs (January
2007).
Stiling, P.D. 1992. Introductory ecology. Prentice Hall, Englewood
Cliffs, New Jersey.
Strahler, A.N. 1952. Dynamic basis of geomorphology. Geological
78
Society of America Bulletin 63: 923-938.
Suttle, K.B., M.E. Power, J.M. Levine and C. McNeely. 2004. How
fine sediment in riverbeds impairs growth and survival of juvenile
salmonids. Ecological Applications 14(4): 969-974.
Swanson, F.J., R.L. Friedriksen and F.M. McCorison. 1982. Material
Transfer in a Western Oregon Forested Watershed, Chapter 8. In
R.L. Edmond, editor. Analysis of Coniferous Forest Ecosystems
in the Western United States. Hutchinson Ross Publishing Co.,
Stroudsbury, PA.
Swift, L.W., G.B. Cunningham, and J.E. Douglass. 1988.
Climatology and hydrology. In W.T. Swank and D.A. Crossley,
Jr., eds. Ecological Studies, Volume 66: Forest Hydrology and
Ecology at Coweeta, pp. 35-55, Springer-Verlag, New York.
Talley, T.S. 2007. Which spatial heterogeneity framework?
Consequences for conclusions about patchy population
distributions. Ecology 88(6): 1476-1489.
Thorson, T.D., S.A. Bryce, D.A. Lammers, A.J. Woods, J.M.
Omernik, J. Kagan, D.E. Pater and J.A. Cornstock. 2003.
Ecoregions of Oregon (Color poster with map, descriptive test,
summary tables, and photographs). U.S. Geological Survey (map
scale 1:1,500,000). Reston, VA.
Thurow, R.F., D.C. Lee and B.E. Rieman. 1997. Distribution and
status of seven native salmonids in the interior Columbia River
basin and Portions of the Klamath River and Great Basins. North
American Journal of Fisheries Management 17: 1094-1110.
Torgersen, C.E., R.E. Gresswell and D.S. Bateman. 2004. Pattern
detection in stream networks: Quantifying spatial variability in
fish distribution. GIS/Spatial Analyses in Fishery and Aquatic
Sciences 405-420.
Trombulak, S.C. and C.A. Frissell. 2000. Review of ecological effects
of roads on terrestrial and aquatic communities. Conservation
Biology 14(1): 18-30.
United States Geological Survey. 2005. NWISWeb: New site for the
nation’s water data. Fact Sheet 128-02.
Vannote, R.L., G.W. Minshall, K.W. Cummins, J.R. Sedell and C.E.
Cushing. 1980. The river continuum concept. Canadian Journal
of Fisheries and Aquatic Science 37: 130-137.
VerHoef, J.M. and K.J. Frost. 2003. A Bayesian hierarchical model
for monitoring harbor seal changes in Prince William Sound,
Alaska. Environmental and Ecological Statistics 10: 201-219.
VerHoef, J.M., E. Peterson and D. Theobald. (2006). Spatial
statistical models that use flow and stream distance. Environmental
and Ecological Statistics 13: 449-464.
Wallington, T.J. 2005. Implications of current ecological thinking
for biodiversity conservation: a review of the salient issues. Ecology
and Society 10(1): 15.
Warren, C.E. and W.J. Liss. 1980. Adaptation to aquatic
environments. Pages 15-40 in R.T. Lackey and L. Nielsen, editors.
Fisheries management. Blackwell Scientific Publications, Oxford,
UK.
Wiens, J.A. 1989. Spatial scaling in ecology. Functional Ecology 3:
385-397.
Flitcroft
Bibliography
Wimberly, M.C., T.A. Spies, C.J. Long and C. Whitlock. 2000.
Simulating historical variability in the amount of old forests in
the Oregon Coast Range. Conservation Biology 14(1): 167-180.
Wright, K.K. and J.L. Li. 2002. From continua to patches:
examining stream community structure over large environmental
gradients. Canadian Journal of Fisheries and Aquatic Sciences 59:
1404-1417.
Wyatt, R.J. 2002. Estimating riverine fish population size from
single- and multiple-pass removal sampling using a hierarchical
model. Canadian Journal of Fisheries and Aquatic Sciences 59:
695-706.
79